π Lecture 37 β Various Economic Applications of Difference Equations
π Overview: This lecture applies first-order difference equations to model dynamic economic phenomena including investment behavior, price dynamics in money markets, fish population management, compound interest calculations, and the Harrod growth model. These applications demonstrate how difference equations provide quantitative frameworks for understanding economic dynamics and equilibrium paths over time.
ποΈ Topics Covered
The lecture covers five economic applications: investment dynamics where investment depends on its previous value, price dynamics in the money market relating inflation to excess money supply, fish stock analysis with population growth and harvesting, compound interest formula derivation from difference equations, and the Harrod model of economic growth including the warranted rate of growth concept.
π Lecture Summary
TOPIC 168: INVESTMENT DYNAMICS USING DIFFERENCE EQUATIONS
Iβ is the amount invested in period t: Iβ = 0.9Iβββ + 60. This is a first-order difference equation where investment depends on 90% of previous period's investment plus a constant term of 60. Given initial investment Iβ = Β£900m, the time-path is found by solving.
Comparing to standard form yβ + a yβββ = c: here a = -0.9 and c = 60. Since a β -1, the definite solution uses formula yβ = (yβ β c/(1+a)) (βa)α΅ + c/(1+a).
π Formula: Iβ = (900 β 60/(1β0.9)) (0.9)α΅ + 60/(1β0.9) β Iβ = 300(0.9)α΅ + 600
The time-path has Iβ (particular integral) = 600 (the equilibrium level) and Iβ (complementary function) = 300(0.9)α΅ (deviation from equilibrium).
π Definition β Dynamic stability: Depends on the base b = 0.9 of exponential term (0.9)α΅.
Since |b| = |0.9| < 1: the time-path is convergent.
Since b = 0.9 > 0: the time-path is non-oscillatory.
π Example: As tββ, (0.9)α΅ β 0, so Iβ β I = 600*. Investment converges to equilibrium of Β£600m.
π‘ Why this matters: The coefficient 0.9 being less than 1 ensures the system moves toward equilibrium over time, representing a stable investment process.
TOPIC 169: PRICE DYNAMICS & MONEY MARKET USING DIFFERENCE EQUATIONS
Money supply Mβ is government-set. Money demand Lβ depends on price level Pβ and real output Y: Lβ = aPβY (a > 0). Inflation (price change) Pβββ β Pβ depends on excess money supply: Pβββ β Pβ = j(Mβ β Lβ) (j > 0).
Substituting demand into the inflation equation: Pβββ β Pβ = j(Mβ β aPβY) β Pβββ β (1 β ajY)Pβ = jMβ
Comparing to yβββ + a yβ = c: here a = β(1 β ajY), c = jMβ, yβ = Pβ.
π Formula: Pβ = (Pβ β Mβ/(aY)) (1 β ajY)α΅ + Mβ/(aY)
The exponential expression is (1 β ajY)α΅.
π Definition β Intertemporal equilibrium price: *P = Mβ/(aY)**, which depends directly on money supply and inversely on the product of the demand parameter a and output Y.
Stability depends on |b| = |1 β ajY|:
- If ajY < 1: |1 β ajY| < 1 β time-path is convergent
- Since b = (1 β ajY) and with ajY < 1: (1 β ajY) > 0 β non-oscillatory
π Example: With parameters such that ajY < 1, the price level converges to equilibrium Mβ/(aY) without oscillations.
TOPIC 170: FISH STOCK ANALYSIS USING DIFFERENCE EQUATION
A fish population increases by 5% each generation, with a catch of 8000 removed from each generation. Let Fβ represent population size for generation t.
(a) The difference equation: Fβ = 1.05Fβββ β 8000
Introducing one generation lag: Fβββ = 1.05Fβ β 8000 β Fβββ β 1.05Fβ = β8000
Comparing to yβββ + a yβ = c: here a = β1.05, c = β8000.
Since a β β1, using solution formula with Fβ = 1000000:
π Formula: Fβ = (1000000 β (β8000)/(1β1.05)) (β(β1.05))α΅ + (β8000)/(1β1.05)
Simplifying: Fβ = (1000000 β 160000)(1.05)α΅ + 160000
Fβ = (β60000)(1.05)α΅ + 160000
Since |b| = |1.05| > 1: time-path is non-convergent (explosive).
Since b = 1.05 > 0: time-path is non-oscillatory.
π Example: As tββ, (1.05)α΅ β β, so Fβ β ββ (the population first declines then becomes negative, indicating extinction at some finite time).
π‘ Why this matters: The 5% growth rate is insufficient to sustain the population against the constant catch of 8000 per generation, demonstrating unsustainable harvesting.
TOPIC 171: INITIAL AMOUNT DEPOSITED IN COMPOUND INTEREST
For compound interest annually: Pβββ = Pβ + iΓPβ where i is interest rate.
π Formula: Pβββ = (1+i)Pβ β Pβββ β (1+i)Pβ = 0
Comparing to yβββ + a yβ = c: here a = β(1+i), c = 0.
Since a β β1 (given i > 0), using solution formula:
π Formula: Pβ = Pβ(1+i)α΅
Stability: |b| = |1+i| > 1 (since i > 0) β time-path is non-convergent (growing).
Pattern: b = (1+i) > 0 β time-path is non-oscillatory.
π Example: If Pβ = 1000 and i = 0.05 (5%), then Pβ = 1000(1.05) = 1050, Pβ = 1000(1.05)Β² = 1102.5, etc.
π‘ Why this matters: This derivation shows the standard compound interest formula emerges naturally from a first-order difference equation, confirming the underlying mathematical structure.
TOPIC 172: HARROD MODEL OF GROWTH
The Harrod model has two key assumptions:
- Sβ = sYβ: saving is constant proportion s of income (MPS = APS)
- Iβ = a(Yβ β Yβββ): investment follows acceleration principle where a is capital-output ratio (COR)
In equilibrium: Iβ = Sβ β a(Yβ β Yβββ) = sYβ
aYβ β aYβββ = sYβ β (s β a)Yβ + aYβββ = 0
Dividing by (s β a): Yβ + (a/(sβa))Yβββ = 0
Introducing lag to match standard form: Yβββ + (a/(sβa))Yβ = 0
Here a = a/(sβa), c = 0.
π Formula: Yβ = Yβ (a/(aβs))α΅
Note the sign change: a/(sβa) = βa/(aβs), so the base becomes a/(aβs).
Since a > 1 (capital-output ratio) and 0 < s < 1 (marginal propensity to save), we have a > s.
|b| = |a/(aβs)|: Since denominator (aβs) < a, we have a/(aβs) > 1 β non-convergent.
Since b = a/(aβs) > 0: non-oscillatory.
π Example: If a = 3, s = 0.2, then Yβ = Yβ (3/(3β0.2))α΅ = Yβ(3/2.8)α΅ = Yβ(1.0714)α΅, showing growth at about 7.14% per period.
TOPIC 173: WARRANTED RATE OF GROWTH
From the Harrod model solution: Yβ = Yβ (a/(aβs))α΅
The warranted rate of growth is the growth rate that maintains equilibrium between saving and investment each year.
Growth rate: G = (Yβ β Yβ)/Yβ
Substituting Yβ = (a/(aβs))Yβ:
G = ((a/(aβs))Yβ β Yβ)/Yβ = a/(aβs) β 1
G = (a β (aβs))/(aβs) = s/(aβs)
π Definition β Warranted growth rate (Gβ): Gβ = s/(aβs), the rate of growth that, if achieved, keeps the economy on an equilibrium path where planned saving equals planned investment.
π Example: With a = 3, s = 0.2: Gβ = 0.2/(3β0.2) = 0.2/2.8 = 0.0714 or 7.14% per period.
π‘ Why this matters: The warranted rate depends on the saving rate s and capital-output ratio a. Any deviation from this path leads to cumulative divergence (instability) in the Harrod model.
β Key Takeaways
All five applications follow the same solution framework for first-order difference equations: identify coefficients a and c, check that a β β1, compute the particular integral (equilibrium) and complementary function (deviation), and analyze dynamic stability using |b| and sign of b. The investment and price dynamics models converge to equilibrium when |b|<1, while the fish stock, compound interest, and Harrod growth models are non-convergent when |b|>1. The money market equilibrium price depends on money supply and output, while the Harrod warranted growth rate Gβ=s/(aβs) represents the knife-edge growth path necessary for continuous equilibrium.
π§ Quick Revision Questions
-
In the investment dynamics problem, what does the value b=0.9 indicate about the time-path's nature and pattern?
-
In the price dynamics model, what condition must hold for the price level to converge to its intertemporal equilibrium?
-
For the fish stock problem with Fβ=1,000,000, growth rate 5%, and catch 8,000 per generation, what happens to the population as tββ?
-
Derive the compound interest formula Pβ=Pβ(1+i)α΅ using the standard first-order difference equation solution method.
-
In the Harrod model, derive the warranted rate of growth Gβ=s/(aβs) and explain why the time-path is non-convergent given a > s.
π Lecture 38 β Cobweb Model and Difference Equations
π Overview: This lecture explains how first-order difference equations model price fluctuations in markets with production lags, such as agriculture. It introduces the cobweb model framework, derives the time-path of price, analyzes convergence and divergence conditions based on relative slopes of demand and supply, and contrasts adaptive expectations in differential versus difference equations.
ποΈ Topics Covered
The lecture covers the framework of the cobweb model using demand and supply functions with time lags, deriving the first-order difference equation for price and its definite solution. It then examines the time-path diagrammatically, analyzing convergent versus divergent oscillatory patterns. A rare case where slopes are equal is considered, followed by a numerical example. Finally, the lecture introduces adaptive expectations and compares their formulation in differential versus difference equations.
π Lecture Summary
TOPIC 174: COBWEB MODEL: FRAMEWORK
Using first-order difference equations, we can understand price fluctuations of products with lags between production and sale, such as agricultural products where planting and harvest occur with a substantial gap.
The Cobweb model is based on demand and supply forces. The demand side has no lags; a buyer is concerned with the current price of agricultural products. Mathematically: Qd(t) = D(Pt), where current demand depends on current price.
The supply side involves a lag; it takes a season (time period) for a crop to get ready. Supply decisions are based on past prices. Mathematically: Qs(t+1) = S(Pt) β future supply depends on current price. Since Qs(t+1) cannot be equated Qd(t) due to different time periods, we convert Qs(t+1) into Qs(t) by introducing a lag (-1): Qs(t) = S(Pt-1).
π Definition β Unlagged Demand Function: Qd(t) = D(Pt) π Definition β Lagged Supply Function: Qs(t) = S(Pt-1) π Formula β Unlagged Demand Equation: Qd(t) = a - Ξ±(Pt) π Formula β Lagged Supply Equation: Qs(t) = -Ξ² + Ξ΄(Pt-1)
For equilibrium in the agricultural market: Qd = Qs. Setting Qd(t) = Qs(t): a - Ξ±(Pt) = -Ξ² + Ξ΄(Pt-1) Rearranging: Ξ±(Pt) + Ξ΄(Pt-1) = a + Ξ² Normalizing by dividing by Ξ±: Pt + (Ξ΄/Ξ±)(Pt-1) = (a + Ξ²)/Ξ±
π Definition β First Order Difference Equation Standard Form: y(t+1) - ay(t) = b
To match the standard form, we introduce a time lead: Pt+1 + (Ξ΄/Ξ±)(Pt) = (a + Ξ²)/Ξ±, where a = Ξ΄/Ξ±, c = (a + Ξ²)/Ξ±, with yt+1 = Pt+1, yt = Pt, and a β -1.
The definite solution formula for a first-order difference equation with a β -1 is: yt = (y0 - c/(1+a))(-a)^t + c/(1+a)
Applying this to the price path: Pt = {P0 - [(a+Ξ²)/Ξ±] / [1 + (Ξ΄/Ξ±)]} (-Ξ΄/Ξ±)^t + [(a+Ξ²)/Ξ±] / [1 + (Ξ΄/Ξ±)] Simplifying: Pt = {P0 - (a+Ξ²)/(Ξ±+Ξ΄)}(-Ξ΄/Ξ±)^t + (a+Ξ²)/(Ξ±+Ξ΄)
π Definition β Stationary Equilibrium Price (P)**: P = (a+Ξ²)/(Ξ±+Ξ΄)
This gives the final time-path solution: Pt = (P0 - P)(-Ξ΄/Ξ±)^t + P**
The term A = (P0 - P)* plays two roles:
- Scale Effect: The absolute value |P0 - P*| affects the magnitude of the complementary function Pc = A(-Ξ΄/Ξ±)^t. If |A|=0, Pc=0; if |A|<1, Pc is minified; if |A|=1, neutral; if |A|>1, amplified.
- Mirror Effect: The sign of (P0 - P*) determines the sign of Pc. If (P0 - P*) > 0, Pc > 0 (retain); if (P0 - P*) < 0, Pc < 0 (reverse).
π‘ Why this matters: The cobweb model shows how past prices influence current supply, creating a dynamic price path that may converge to or diverge from equilibrium.
TOPIC 175: COBWEB MODEL: DIAGRAMMATICALLY
The time-path Pt = (P0 - P*)(-Ξ΄/Ξ±)^t + P* contains the exponential expression (-Ξ΄/Ξ±)^t which determines the nature and pattern of the time path.
π Definition β Nature: Whether the time-path converges or diverges, determined by |b| = | -Ξ΄/Ξ± | β 1. π Definition β Pattern: Whether the time-path oscillates or not, determined by b = (-Ξ΄/Ξ±) β 0.
Since both Ξ΄ and Ξ± are positive (Ξ΄, Ξ± > 0), the base b = -Ξ΄/Ξ± is always negative, meaning the pattern is always oscillatory.
The nature depends on relative slopes:
- |b| < 1 (Ξ΄ < Ξ±): Convergent β slope of supply is less than slope of demand. This produces damped oscillations where the price path converges toward equilibrium while oscillating above and below it.
- |b| > 1 (Ξ΄ > Ξ±): Non-convergent (Divergent) β slope of supply is greater than slope of demand. This produces explosive oscillations where the price path diverges away from equilibrium while oscillating.
The left panels show the converging/diverging dynamics in the market with supply and demand curves. The right panels plot the time-path of price (Pt) oscillating around equilibrium price (pe).
π Example: Convergent case shows the price oscillating with decreasing amplitude toward equilibrium. Divergent case shows increasing amplitude away from equilibrium.
Conclusion: The nature of the time-path is primarily dependent on the relative slope of demand and supply curves (Ξ΄ β Ξ±).
TOPIC 176: COBWEB MODEL: RARE CASE
The rare case occurs when Ξ΄ = Ξ±, meaning the slopes of supply and demand are equal.
When Ξ΄ = Ξ±, the base b = -Ξ΄/Ξ± = -1.
- |b| = |-1| = 1 (neither convergent nor divergent): Non-convergent
- b = -1 < 0: Oscillatory
This produces uniform oscillations where the price uniformly oscillates above and below the equilibrium price by a constant amplitude.
The left panel shows equi-distant dynamics in the market (Ξ΄ = Ξ±). The right panel plots the time-path of price (Pt) uniformly oscillating around equilibrium price (pe).
Conclusion: The existence of the uniform oscillation case is a rare event as the equality of slopes is a matter of chance (Ξ΄ = Ξ±). Any slight inequality can lead to convergence or divergence cases.
TOPIC 177: COBWEB MODEL: NUMERICAL
Given: Qdt = 18 - 3Pt and Qst = -3 + 4Pt-1, find the time-path of price.
Solution: Here, a = 18, Ξ± = 3, Ξ² = 3, Ξ΄ = 4.
For equilibrium: Qd(t) = Qs(t) 18 - 3Pt = -3 + 4Pt-1 -3Pt - 4Pt-1 = -21 3Pt + 4Pt-1 = 21 Normalizing by dividing by 3: Pt + (4/3)Pt-1 = 7
Introducing time lead: Pt+1 + (4/3)Pt = 7 Comparing with standard form: a = Ξ΄/Ξ± = 4/3, c = 7
Since a β -1, using the definite solution formula: Pt = {P0 - c/(1+a)}(-a)^t + c/(1+a) Pt = {P0 - 7 / [1 + (4/3)]}(-4/3)^t + 7 / [1 + (4/3)] Pt = {P0 - 7 / (7/3)}(-4/3)^t + 7 / (7/3) Pt = {P0 - 3}(-4/3)^t + 3
π Example: The time-path is Pt = (P0 - 3)(-4/3)^t + 3
Analysis:
- |b| = | -4/3 | = 1.33 > 1: Non-convergent (Divergent)
- b = (-4/3) < 0: Oscillatory
Conclusion: The time-path is divergent in nature and oscillatory in pattern β dynamically unstable.
TOPIC 178: DIFFERENCE EQUATIONS AND ADAPTIVE EXPECTATIONS
Consider a market model with adaptive expectations:
- Qdt = Qst
- Qdt = -a + Ξ±Pt
- Qst = Ξ² - Ξ΄P*t
Where P*t is the expected price in the current time period t.
π Definition β Adaptive Expectations: Pt = Pt-1 + Ξ·(Pt-1 - P*t-1) where 0 < Ξ· β€ 1. The expectation in the current period equals the previous expectation plus a fraction Ξ· of the discrepancy between actual and expected price in the previous period.
Effect of Adaptive Price Expectations:
- If (Pt-1 > P*t-1) β price in previous period was greater than expected: under-expectation. Expectations are revised upward.
- If (Pt-1 < P*t-1) β price in previous period was less than expected: over-expectation. Expectations are revised downward.
π Example: If Pt-1 = 20, Pt-1 = 10, Ξ· = 0.5, then Pt = 20 + 0.5(20-10) = 25 (revised upward). If Pt-1 = 20, Pt-1 = 30, Ξ· = 0.5, then Pt = 20 + 0.5(20-30) = 15 (revised downward).
Role of Ξ· = 1: When Ξ· = 1, the expectation equation becomes: Pt = Pt-1 + 1(Pt-1 - Pt-1) Pt = Pt-1
This means the seller expects the price this year to be the same as last year β the situation observed in the cobweb model.
π Definition β Adaptive Expectations in Differential vs. Difference Equations:
| Feature | Differential Equations | Difference Equations |
|---|---|---|
| Equation | dΟ/dt = j(p - Ο) | Pt = Pt-1 + Ξ·(Pt-1 - P*t-1) |
| Change | dΟ/dt | Pt - Pt-1 |
| Adjustment coefficient | j | Ξ· |
| Discrepancy term | (p - Ο) | (Pt-1 - P*t-1) |
Observation: Both differential and difference equations have the same structure when including adaptive expectations. The adjustment coefficient j in differentials corresponds to Ξ· in difference equations, and the discrepancy term (p-Ο) in differentials corresponds to (Pt-1 - P*t-1) in difference equations.
β Key Takeaways
The cobweb model uses first-order difference equations to model price fluctuations in markets with production lags. The time-path solution is Pt = (P0 - P*)(-Ξ΄/Ξ±)^t + P*, where convergence occurs when Ξ΄ < Ξ± (supply slope flatter than demand) producing damped oscillations, and divergence when Ξ΄ > Ξ± (supply slope steeper) producing explosive oscillations. The rare case Ξ΄ = Ξ± produces uniform oscillations. In a numerical example with Qd = 18-3Pt and Qs = -3+4Pt-1, the time-path Pt = (P0-3)(-4/3)^t+3 shows divergent oscillations. Finally, adaptive expectations (Pt = Pt-1 + Ξ·(Pt-1 - P*t-1)) generalize the cobweb model and have the same structure in both differential and difference equations, with Ξ·=1 reducing to the simple cobweb case where current expectations equal last period's actual price.
π§ Quick Revision Questions
-
Derive the cobweb model equilibrium condition from Qd(t) = a - Ξ±Pt and Qs(t) = -Ξ² + Ξ΄Pt-1, and show how it becomes a first-order difference equation in Pt.
-
Given the solution Pt = (P0 - P*)(-Ξ΄/Ξ±)^t + P*, explain how the relative values of Ξ΄ and Ξ± determine whether the time-path is convergent, divergent, or uniformly oscillatory.
-
In the numerical example with Qd = 18 - 3Pt and Qs = -3 + 4Pt-1, compute the equilibrium price P* and determine whether the time-path is convergent or divergent.
-
Explain the difference between the scale effect and mirror effect of the coefficient A = (P0 - P*) on the time-path of price.
-
Write the adaptive expectations equation and explain how setting Ξ· = 1 simplifies it to the cobweb model's assumption about price expectations.
π Lecture 39 β A Market Model with Inventory Using Difference Equations
π Overview: This lecture develops a dynamic market model that incorporates inventory, moving beyond the simple Cobweb model which assumes instantaneous market clearing. It introduces a price adjustment mechanism driven by inventory changes, derives the resulting first-order difference equation for the price path, and classifies the possible time paths based on parameter values. This model is crucial for understanding real-world market dynamics where goods can be stored.
ποΈ Topics Covered
The lecture begins by establishing the framework for a market model with inventory, outlining its three key premises. It then derives the solution as a first-order difference equation and obtains the definite time path for price. The core of the lecture is a detailed classification of seven possible time paths based on the value of the characteristic root, with a focus on dynamic stability. Finally, the types of time paths are illustrated graphically and with a numerical example.
π Lecture Summary
TOPIC 179: A MARKET MODEL WITH INVENTORY: FRAMEWORK
The Cobweb model assumes a perishable or non-storable product, leading to market clearance each period. However, many real-world products can be stored, necessitating a model that includes inventory.
The model is built on three premises:
Premise I: The current quantity demanded (Q<sub>dt</sub>) and quantity supplied (Q<sub>st</sub>) are unlagged linear functions of the current price (P<sub>t</sub>).
- Q<sub>dt</sub> = a - Ξ±P<sub>t</sub> (a, Ξ± > 0)
- Q<sub>st</sub> = -Ξ² + Ξ΄P<sub>t</sub> (Ξ², Ξ΄ > 0)
Premise II: Price adjustment is not driven by market clearance but by sellers considering the inventory situation. An accumulated inventory from the previous period leads sellers to drop the current price to move the merchandise. Conversely, a decumulated inventory raises the current price. The price adjustment equation is:
- P<sub>t+1</sub> = P<sub>t</sub> - Ο(Q<sub>st</sub> - Q<sub>dt</sub>) (Ο > 0)
Premise III: The price adjustment from one period to the next is inversely proportional to the observed change in inventory (stock).
Mathematizing the premises:
- Ο is the stock-induced-price-adjustment coefficient.
- The term (Q<sub>st</sub> - Q<sub>dt</sub>) represents the inventory (stock), while (Q<sub>dt</sub> - Q<sub>st</sub>) is excess demand. π‘ Why this matters: This framework directly links price changes to inventory levels, providing a more realistic model of price determination than the Cobweb model.
TOPIC 180: A MARKET MODEL WITH INVENTORY: SOLUTION
To find the time path of price P<sub>t</sub>, we substitute the demand and supply functions into the price-adjustment function:
- P<sub>t+1</sub> = P<sub>t</sub> - Ο[(-Ξ² + Ξ΄P<sub>t</sub>) - (a - Ξ±P<sub>t</sub>)]
- Simplifying: P<sub>t+1</sub> - [1 - Ο(Ξ± + Ξ΄)]P<sub>t</sub> = Ο(a + Ξ²)
This is a First Order Difference Equation representing the market model with inventory. Comparing it to the standard form y<sub>t+1</sub> + ay<sub>t</sub> = c:
- a = -[1 - Ο(Ξ± + Ξ΄)]
- c = Ο(a + Ξ²)
- y<sub>t+1</sub> = P<sub>t+1</sub>
- y<sub>t</sub> = P<sub>t</sub>
Using the formula for the definite solution of a first-order difference equation (assuming a β -1):
- P<sub>t</sub> = (P<sub>0</sub> - P) [1 - Ο(Ξ± + Ξ΄)]<sup>t</sup> + P**
Here, P* = (a + Ξ²)/(Ξ± + Ξ΄) is the stationary equilibrium or intertemporal equilibrium price. The component (P<sub>0</sub> - P)* represents the deviation from equilibrium at time 0. The term b = [1 - Ο(Ξ± + Ξ΄)] is the characteristic root that determines the dynamic behavior of the price path.
π Definition β *Stationary Equilibrium (P)**: The long-run equilibrium price towards which the market price converges if the time path is convergent. It is the price where the difference equation stabilizes.
π Formula: P<sub>t</sub> = (P<sub>0</sub> - P)b<sup>t</sup> + P** β The time path of price is a function of the initial deviation from equilibrium, the characteristic root 'b' raised to the power of 't', and the stationary equilibrium price.
TOPIC 181: A MARKET MODEL WITH INVENTORY: TYPES OF TIME PATH
The time path's nature and pattern are determined by the characteristic root b = [1 - Ο(Ξ± + Ξ΄)]. Seven theoretical regions exist, though two are ruled out by the parametric restriction Ο > 0. The dynamic stability depends on the absolute value and sign of 'b'.
| Region | Value of 'b' | Pattern | Condition for Ο (given Ξ±, Ξ΄ > 0) | Stability & Pattern |
|---|---|---|---|---|
| I | b > 1 | Non-oscillatory | Ο < 0 (violates Ο > 0) | Ruled out |
| II | b = 1 | Non-oscillatory | Ο = 0 (violates Ο > 0) | Ruled out |
| III | 0 < b < 1 | Non-oscillatory | 0 < Ο < 1/(Ξ± + Ξ΄) | Convergent, Non-oscillatory |
| IV | b = 0 | Non-oscillatory | Ο = 1/(Ξ± + Ξ΄) | Pre-convergent (converges in one step), Non-oscillatory |
| V | -1 < b < 0 | Oscillatory | 1/(Ξ± + Ξ΄) < Ο < 2/(Ξ± + Ξ΄) | Convergent, Oscillatory (damped oscillations) |
| VI | b = -1 | Oscillatory | Ο = 2/(Ξ± + Ξ΄) | Non-convergent, Oscillatory (uniform oscillations) |
| VII | b < -1 | Oscillatory | Ο > 2/(Ξ± + Ξ΄) | Divergent, Oscillatory (explosive oscillations) |
Conclusion: Out of the seven theoretical cases, only five (III to VII) are feasible given the constraint Ο > 0. The stability of the market model with inventory is entirely governed by the magnitude of the stock-induced-price-adjustment coefficient (Ο) relative to the sum of the slopes of the demand and supply curves (Ξ± + Ξ΄).
TOPIC 182: A MARKET MODEL WITH INVENTORY: GRAPHICALLY
The five feasible regions (III to VII) can be plotted on a graph with Ο (the stock-induced-price-adjustment coefficient) on the vertical axis and (Ξ± + Ξ΄) (the sum of demand and supply curve slopes) on the horizontal axis. Only the first quadrant is relevant as both variables are positive.
The graph maps the conditions from the table above into distinct regions. The lines Ο = 1/(Ξ± + Ξ΄) and Ο = 2/(Ξ± + Ξ΄) are rectangular hyperbolas that serve as boundaries between the different stability regimes. As Ο increases relative to (Ξ± + Ξ΄), the time path moves from convergent non-oscillatory (Region III) to divergent oscillatory (Region VII). π‘ Why this matters: This graph provides a powerful visual tool for determining the dynamic stability of a market with inventory for any given set of parameter values (Ξ±, Ξ΄, Ο).
TOPIC 183: A MARKET MODEL WITH INVENTORY: NUMERICAL
Example #1: Given the demand, supply, and price-adjustment functions:
- Q<sub>dt</sub> = 21 - 2P<sub>t</sub> (a=21, Ξ±=2)
- Q<sub>st</sub> = -3 + 6P<sub>t</sub> (Ξ²=3, Ξ΄=6)
- P<sub>t+1</sub> = P<sub>t</sub> - 0.3(Q<sub>st</sub> - Q<sub>dt</sub>) (Ο=0.3)
Find the time path P<sub>t</sub> and determine its dynamic stability.
Solution:
-
Find the difference equation:
- P<sub>t+1</sub> = P<sub>t</sub> - 0.3[(-3 + 6P<sub>t</sub>) - (21 - 2P<sub>t</sub>)]
- Simplifying: P<sub>t+1</sub> + 1.4P<sub>t</sub> = 7.2
-
Solve for the definite time path:
- Using the formula P<sub>t</sub> = (P<sub>0</sub> - P)(-1.4)<sup>t</sup> + P**, where P = (a + Ξ²)/(Ξ± + Ξ΄) = (21 + 3)/(2 + 6) = 3*.
- P<sub>t</sub> = (P<sub>0</sub> - 3)(-1.4)<sup>t</sup> + 3 (This is the definite time path)
-
Determine dynamic stability using 'b':
- Here, b = 1 - Ο(Ξ± + Ξ΄) = 1 - 0.3(2 + 6) = 1 - 2.4 = -1.4
- Check the absolute value: |b| = |-1.4| = 1.4 > 1
- Check the sign: b = -1.4 < 0
- Conclusion: The time path is Divergent (Unstable) and Oscillatory.
- This places the model in Region VII.
-
Verify dynamic stability using parameters Ο, Ξ±, Ξ΄:
- The condition for Region VII (Divergent & Oscillatory) is: Ο > 2/(Ξ± + Ξ΄)
- Here, Ο = 0.3 and 2/(Ξ± + Ξ΄) = 2/(2 + 6) = 0.25
- Since 0.3 > 0.25, we reach the same conclusion: Divergent & Oscillatory.
β Key Takeaways
The lecture presents a more realistic market model where price adjusts based on inventory levels, unlike the instant-clearing Cobweb model. The core dynamic equation, P<sub>t+1</sub> - [1 - Ο(Ξ± + Ξ΄)]P<sub>t</sub> = Ο(a + Ξ²), and its solution, P<sub>t</sub> = (P<sub>0</sub> - P)b<sup>t</sup> + P**, are essential for understanding price evolution. The characteristic root b = [1 - Ο(Ξ± + Ξ΄)] is the sole determinant of the time path's stability and oscillatory nature, leading to five feasible regions. Dynamic stability requires |b| < 1, which translates to a small Ο relative to (Ξ± + Ξ΄), specifically Ο < 2/(Ξ± + Ξ΄) for oscillatory convergence. The numerical example demonstrates how to derive the time path and classify its stability using both 'b' and the parameters directly.
π§ Quick Revision Questions
- What are the three key premises of the market model with inventory, and how do they differ from the premises of the Cobweb model?
- Derive the first-order difference equation for price in this model, starting from the functions Q<sub>dt</sub> = a - Ξ±P<sub>t</sub>, Q<sub>st</sub> = -Ξ² + Ξ΄P<sub>t</sub>, and P<sub>t+1</sub> = P<sub>t</sub> - Ο(Q<sub>st</sub> - Q<sub>dt</sub>). Identify the characteristic root 'b'.
- State the conditions on Ο, Ξ±, and Ξ΄ for the time path to be: (a) convergent and oscillatory, and (b) divergent and oscillatory.
- In the numerical example, what is the intertemporal equilibrium price P*, and why is the time path classified as divergent and oscillatory?
- Explain how the graphical representation with Ο and (Ξ± + Ξ΄) helps in determining the dynamic stability of this market model.
π Lecture 40 β SECOND ORDER DIFFERENCE EQUATIONS
π Overview: This lecture introduces second-order difference equations, focusing on how to find the particular integral (yp) and the complementary function (yc) for these equations. It covers the standard form, homogenous and non-homogenous cases, and the three subcases for finding particular solutions. The lecture also explains how to find the complementary function for distinct real roots, repeated real roots, and complex roots, which are essential for solving economic models involving intertemporal equilibrium.
ποΈ Topics Covered
This lecture covers the standard form of second-order difference equations and the identification of homogenous and non-homogenous cases based on the constant term. It details the three subcases for finding the particular integral (yp) based on the values of coefficients, including the caveats for each subcase. The lecture then explains the complementary function (yc), adapting concepts from differential equations and solving through the characteristic roots, covering distinct real roots, repeated real roots, and complex roots with numerical examples.
π Lecture Summary
TOPIC 184: SECOND-ORDER DIFFERENCE EQUATIONS: PARTICULAR INTEGRAL
Similar to second-order differential equations, difference equations can also acquire a second order of difference (recurrence). The second difference of yt, denoted as βΒ²yt, is defined as β(βyt). Expanding this, βΒ²yt = yt+2 - 2yt+1 + yt.
The standard form of a linear, non-homogeneous second-order difference equation with constant coefficients (a1, a2) and constant term (c) is: yt+2 + a1yt+1 + a2yt = c.
The order is 2nd (due to the 2nd difference), the degree is linear/first degree, and the coefficients are constant. The course deals only with linear cases and does not include cross-terms.
Based on the value of c, there are two main cases: Homogenous Case (c = 0, yp = 0) and Non-Homogenous Case (c β 0, yp β 0). Both cases are further divided into three sub-cases based on the values of a1 and a2.
π Definition β Particular Integral (yp): Any particular solution (containing no arbitrary constants) of the complete equation. yp provides the equilibrium value of yt in an intertemporal sense. Formula: The general solution is yt = yc + yp, where yc is the deviation from equilibrium and yp is the equilibrium value.
Subcase β I: Since yp can be any solution, we first try the simplest, y = k (a constant). Substituting yt = yt+1 = yt+2 = k into the general form yt+2 + a1yt+1 + a2yt = c gives: k + a1k + a2k = c β k(1 + a1 + a2) = c. π Formula: yp = c / (1 + a1 + a2), provided that a1 + a2 β -1.
Subcase β II: If a1 + a2 = -1, the previous formula gives an undefined answer. We then try a non-constant solution, y = kt. Substituting yt = kt, yt+1 = k(t+1), and yt+2 = k(t+2) into the equation and simplifying, using the condition a1 + a2 = -1, we get k[a1 + 2] = c. π Formula: yp = (c / (a1 + 2)) * t, provided that a1 β -2.
Subcase β III: If a1 = -2 (and already a1 + a2 = -1), both previous subcase formulas fail. We then try y = ktΒ². Substituting yt = ktΒ², yt+1 = k(t+1)Β², and yt+2 = k(t+2)Β² into the equation and simplifying, using a1 + a2 = -1 and a1 = -2, we get k[2] = c. π Formula: yp = (c / 2) * tΒ². π‘ Why this matters: This hierarchy of "tries" (constant, then t, then tΒ²) ensures we always find a valid particular integral for any non-homogenous second-order difference equation with constant coefficients.
TOPIC 185: SECOND-ORDER DIFFERENCE EQUATIONS: YP NUMERICALS
Non-Homogenous Subcase β I: c β 0, a1 + a2 β -1. π Example: Solve yt+2 - 3yt+1 + 4yt = 12. Here, a1 = -3, a2 = 4, c = 12. Check: a1 + a2 = 1 β -1. Using formula yp = c / (1 + a1 + a2) = 12 / (1 + (-3) + 4) = 12 / 2 = 6.
Non-Homogenous Subcase β II: c β 0, a1 + a2 = -1, a1 β -2. π Example: Solve yt+2 + yt+1 - 2yt = 12. Here, a1 = 1, a2 = -2, c = 12. Check: a1 + a2 = -1, a1 = 1 β -2. Using formula yp = (c / (a1+2)) * t = (12 / (1+2)) * t = 4t.
Non-Homogenous Subcase β III: c β 0, a1 + a2 = -1, a1 = -2. π Example: Solve yt+2 - 2yt+1 + yt = 12. Here, a1 = -2, a2 = 1, c = 12. Check: a1 + a2 = -1, a1 = -2. Using formula yp = (c/2) * tΒ² = (12/2) * tΒ² = 6tΒ².
The summary tables show that for the Homogenous Case (c = 0), the particular integral is yp = 0 for all subcases. For the Non-Homogenous Case (c β 0), the yp values are as calculated above.
TOPIC 186: SECOND-ORDER DIFFERENCE EQUATIONS: YC DISTINCT REAL ROOTS
The Complementary Function (yc) is the general solution (containing arbitrary constants) of the reduced equation. yc provides the deviation of the time path y(t) from its equilibrium value at every point in time. The solution is y(t) = yc + yp.
The form of yc for difference equations is adapted from differential equations, replacing natural exponentials with exponential functions based on roots (b). For second-order equations with distinct real roots (b1 β b2): π Formula: yc = A1(b1)^t + A2(b2)^t.
Nature of Roots: The discriminant (a1Β² - 4a2) determines the nature. If a1Β² - 4a2 > 0, roots are distinct real roots. The roots (b1, b2) are found using the quadratic formula. π Formula: b1, b2 = (-a1 Β± β(a1Β² - 4a2)) / 2.
π Example: Solve yt+2 + yt+1 - 2yt = 12, given y0 = 4, y1 = 5. From Topic 185, yp = 4t. Calculate roots: a1 = 1, a2 = -2 β a1Β² - 4a2 = 1 - 4(-2) = 9 > 0. Roots: b1, b2 = (-1 Β± β9)/2 = 1 and -2. yc = A1(1)^t + A2(-2)^t. General solution: y(t) = A1 + A2(-2)^t + 4t. Definite Solution: Use initial conditions. For t=0: y0 = 4 = A1 + A2(1) + 0 β A1 + A2 = 4. For t=1: y1 = 5 = A1 + A2(-2) + 4 β A1 - 2A2 = 1. Solving: A2 = 1, A1 = 3. Definite solution: y(t) = 3 + (-2)^t + 4t.
TOPIC 187: SECOND-ORDER DIFFERENCE EQUATIONS: YC REPEATED REAL ROOTS
For second-order equations with repeated real roots (b1 = b2 = b): π Formula: yc = A3(b)^t + A4 * t * (b)^t.
Nature of Roots: If a1Β² - 4a2 = 0, roots are repeated real roots.
π Example: Solve yt+2 + 6yt+1 + 9yt = 4. Here, a1 = 6, a2 = 9, c = 4. Check: a1 + a2 = 15 β -1, so Subcase I applies. yp = c / (1 + a1 + a2) = 4 / (1 + 6 + 9) = 4/16 = 1/4. Calculate roots: a1Β² - 4a2 = 36 - 36 = 0. Root: b = -a1/2 = -6/2 = -3. yc = A3(-3)^t + A4 * t * (-3)^t. General solution: y(t) = A3(-3)^t + A4 * t * (-3)^t + 1/4. π‘ Why this matters: The inclusion of the "t" factor in the repeated roots case changes the dynamics of the time path, leading to a different convergence behavior compared to distinct roots.
TOPIC 188: SECOND-ORDER DIFFERENCE EQUATIONS: YC COMPLEX ROOTS
For second-order equations with complex roots (h Β± ui): π Formula: yc = R^t (A5 cos ΞΈt + A6 sin ΞΈt).
Nature of Roots: If a1Β² - 4a2 < 0, roots are complex roots. Here, h = -a1/2 and u = β(4a2 - a1Β²)/2 are defined. The modulus R (or amplitude) is calculated as R = β(hΒ² + uΒ²), and the angle ΞΈ is found via sin ΞΈ = u/R or cos ΞΈ = h/R.
π Example: Solve yt+2 + 0*yt+1 + (1/4)yt = 5. (Note: "self generated" in text indicates a1=0) Here, a1 = 0, a2 = 1/4, c = 5. Check: a1 + a2 = 1/4 β -1, so Subcase I applies. yp = c / (1 + a1 + a2) = 5 / (1 + 0 + 1/4) = 5 / (5/4) = 4. Calculate roots: a1Β² - 4a2 = 0 - 4(1/4) = -1 < 0. h = -a1/2 = 0, u = β(4a2 - a1Β²)/2 = β(1 - 0)/2 = 1/2. R = β(hΒ² + uΒ²) = β(0 + (1/2)Β²) = 1/2. ΞΈ: sin ΞΈ = u/R = (1/2) / (1/2) = 1 β ΞΈ = 90Β° (or Ο/2 radians). yc = (1/2)^t (A5 cos(Ο/2 * t) + A6 sin(Ο/2 * t)). General solution: yt = (1/2)^t {A5 cos(Ο/2 * t) + A6 sin(Ο/2 * t)} + 4.
β Key Takeaways
The core of solving second-order difference equations is decomposing the solution into a particular integral (yp) for the equilibrium and a complementary function (yc) for deviations from equilibrium. The formula for yp depends on three subcases, determined by the coefficients a1 and a2: if a1 + a2 β -1, use the constant formula; if not, and a1 β -2, use the t formula; otherwise, use the tΒ² formula. The form of yc is determined by the nature of the roots from the characteristic equation, leading to distinct real, repeated real, or complex root solutions, which dictate the dynamic behavior of the time path. Finally, initial conditions are used to find the definite solution by solving for the arbitrary constants in the general solution.
π§ Quick Revision Questions
- What is the standard form of a second-order linear difference equation with constant coefficients?
- What are the three subcases for finding the particular integral (yp), and what conditions define each subcase?
- How do you determine the nature of the roots (distinct real, repeated real, complex) for a second-order difference equation?
- What is the formula for the complementary function (yc) in the case of repeated real roots?
- In the complex roots case, how are the modulus R and the angle ΞΈ defined and used in the yc formula?
π Lecture 41 β Dynamic Stability: Distinct Real Roots
π Overview: This lecture explores the concept of dynamic stability in difference equations, focusing on the conditions under which a system converges to equilibrium over time. It covers three distinct cases: distinct real roots, repeated real roots, and complex roots, each with its own stability criteria. Understanding these stability conditions is crucial for predicting the long-run behavior of economic models.
ποΈ Topics Covered
The lecture covers dynamic stability for distinct real roots, repeated real roots, and complex roots, with numerical examples for each case. It explains how the absolute values of the roots determine whether the complementary function (deviation from equilibrium) converges to zero, leading to stability, or diverges, leading to instability. The "hair splitting" approach is used to analyze the behavior of the exponential terms.
π Lecture Summary
TOPIC 189: DYNAMIC STABILITY: DISTINCT REAL ROOTS
For the distinct real roots case, the general solution of a difference equation is given by:
(y_t = A_1(b_1)^t + A_2(b_2)^t + y_p)
where the complementary function, (y_c = A_1(b_1)^t + A_2(b_2)^t), represents the deviation from the equilibrium (y_p). (A_1) and (A_2) are constants determined by initial conditions, and (t) is the independent variable spanning ((0, \infty)). The dynamic stability of the system depends entirely on the roots (b_1) and (b_2).
π Definition β Dynamic Stability: A system is dynamically stable if, over time, the complementary function ((y_c)) converges to zero, meaning the deviation from equilibrium fades away. This occurs when both roots ((b_1) and (b_2)) are negative, causing the exponential terms to decay.
The "hair splitting" analysis shows that if both (b_1) and (b_2) are negative, the exponential terms (A_1(b_1)^t) and (A_2(b_2)^t) will fade away due to exponential decay. Consequently, their sum, (y_c), also fades away. A caveat is that both (b_1) and (b_2) must be negative; otherwise, convergence may not occur.
π Formula: Dynamic stability condition for distinct real roots: (|b_1| < 1) and (|b_2| < 1).
π Example: The lecture provides numerical examples to illustrate the concept.
- Example 1: (y(t) = 3(1)^t + (2)^t + 4t). Here, (|b_1| = 1 \not< 1) and (|b_2| = 2 \not< 1). The first term remains constant, and the second term increases; the deviation increases, leading to Divergence and Dynamic Instability.
- Example 2: (y(t) = 3(0.5)^t + (-2)^t + 4t). Here, (|b_1| = 0.5 < 1) but (|b_2| = 2 \not< 1). The first term decreases, but the second term increases; the deviation increases, leading to Divergence and Dynamic Instability.
- Example 3: (y(t) = 3(-0.8)^t + (0.1)^t + 4t). Here, (|b_1| = 0.8 < 1) and (|b_2| = 0.1 < 1). Both terms decrease; the deviation decreases, leading to Convergence and Dynamic Stability.
This confirms that for dynamic stability, the absolute value of both roots must be less than 1: ((|b_1| \text{ and } |b_2|) < 1).
TOPIC 190: DYNAMIC STABILITY: REPEATED REAL ROOTS
For the repeated real roots case, the solution is:
(y_t = A_3(b)^t + A_4 t (b)^t + y_p)
The complementary function is (y_c = A_3(b)^t + A_4 t (b)^t). The deviation from equilibrium depends on the single repeated root (b). (A_3) and (A_4) are constants.
π Definition β Repeated Real Roots Case: A condition in difference equations where the characteristic equation has a single real root with multiplicity, requiring a modified solution form that includes a term multiplied by (t).
The "hair splitting" analysis shows that if (b) is negative, the exponential terms (A_3(b)^t) and (A_4 t(b)^t) will fade away. Although the second term has an additional (t) multiplier, it does not reverse the dynamics of the exponential decay. The sum of the two terms will also converge to zero.
π Formula: Dynamic stability condition for repeated real roots: (|b| < 1).
π Example: Numerical examples are provided.
- Example 1: (y = 2(3)^t + t(3)^t + \frac{1}{4}t). Here, (|b| = 3 \not< 1). Both exponential terms increase, the deviation increases, and the system is Dynamically Unstable (Divergence).
- Example 2: (y = 2(0.3)^t + t(0.3)^t + \frac{1}{4}t). Here, (|b| = 0.3 < 1). Both exponential terms decrease, the deviation decreases, and the system is Dynamically Stable (Convergence).
This confirms that for dynamic stability, the absolute value of the repeated root (b) must be less than 1: (|b| < 1).
TOPIC 191: DYNAMIC STABILITY: COMPLEX ROOTS
For the complex roots case, the solution involves trigonometric functions:
(y_t = R^t (A_5 \cos \theta t + A_6 \sin \theta t) + y_p)
The complementary function is (y_c = R^t (A_5 \cos \theta t + A_6 \sin \theta t)). The deviation from equilibrium depends on the modulus (R) and the angle (\theta) of the complex roots.
π Definition β Complex Roots Case: A condition in difference equations where the characteristic equation has a pair of complex conjugate roots, leading to a solution that involves sinusoidal functions and a multiplicative factor (R^t).
The "hair splitting" analysis shows that the multiplicand ((A_5 \cos \theta t + A_6 \sin \theta t)) produces a "stepped fluctuation" with a wave-like graph of uniform amplitude. Convergence depends on the multiplier (R^t). If (R < 1), then (R^t \rightarrow 0) as (t \rightarrow \infty), damping the fluctuations and leading to convergence.
π Formula: Dynamic stability condition for complex roots: (R < 1).
π Example: The lecture provides three numerical examples.
- Example 1: (y_t = (2)^t { \cos(1)t + 3 \sin(1)t } + 4). Here, the modulus (R = 2 \not< 1). The multiplier (R^t) increases, creating an Explosive Fluctuation and Divergence; the system is Dynamically Unstable.
- Example 2: (y_t = (1)^t { \cos(1)t + 3 \sin(1)t } + 4). Here, the modulus (R = 1). The multiplier (R^t) is constant, creating a Uniform Fluctuation that is neither divergent nor convergent; the system is Neither Dynamically Stable nor Unstable.
- Example 3: (y_t = (\frac{1}{2})^t { \cos(1)t + 3 \sin(1)t } + 4). Here, the modulus (R = 0.5 < 1). The multiplier (R^t) decreases, creating a Damped Fluctuation and Convergence; the system is Dynamically Stable.
This confirms that for dynamic stability in the complex root case, the condition (R < 1) must hold.
β Key Takeaways
A student must remember that dynamic stability for all three cases depends on the absolute value (or modulus for complex roots) of the characteristic roots. For distinct real roots, both (|b_1|) and (|b_2|) must be less than 1 for stability. For repeated real roots, (|b|) must be less than 1. For complex roots, the modulus (R) must be less than 1. The complementary function (y_c), which represents the deviation from equilibrium, must converge to zero as time (t) approaches infinity. If any root's absolute value or modulus is greater than or equal to 1, the system will exhibit divergence or non-convergent fluctuations, making it dynamically unstable. The numerical examples confirm that the "exponential decay" of terms like ((b)^t) is the core mechanism driving convergence when (|b|<1).
π§ Quick Revision Questions
- What is the general condition for dynamic stability when a difference equation has two distinct real roots?
- In the repeated real roots case, why does the additional 't' factor in the term (A_4 t (b)^t) not prevent dynamic stability if (|b|<1)?
- For a system with complex roots, what does it mean if the modulus (R) is exactly 1?
- In the distinct real roots numerical example, why does the system (y(t) = 3(0.5)^t + (-2)^t + 4t) diverge despite (|0.5|<1)?
- What is the key difference in the solution form between the distinct real roots case and the repeated real roots case?