Project 6:
Chaos in a financial model
Simon Wu, Ziyun Zeng, Michael Li, Jiaye Zhang, Haochen Li, Yixin
Ma
Agenda
• Introduction of the model
• Stability of the balance points of the model
• Graphs
• Time delay model
Introduction of the financial model
• π₯ =π§+ π¦−π π₯
• π¦ = 1 − ππ¦ − π₯ 2
• π§ = −π₯ − ππ§
The economic meaning of variables and parameters:
• x: interest rate
• y: the investment demand
• z: price index
π₯ =π§+ π¦−π π₯
• π₯ = π1 π¦ − π π₯ + π2 π§
• π1 , π2 are constants and a is the amount saving.
What does this equation mean in reality?
• change of interest rate is influenced by 2 parts, i.e. price index
and contradictions from the investment market.
2
π¦ = 1 − ππ¦ − π₯
• π¦ = π3 π − ππ¦ − πΌπ₯ 2
• π3 , b, πΌ are constants, r is the natural growth rate
• What does this equation mean in reality?
• Change of investment demand is affected by parts, i.e.
investment demand and interest rate.
π§ = −π₯ − ππ§
• π§ = −π4 π₯ − ππ§
• π4 is constant and c is the elasticity of demand of commercial
market
• What does this equation mean in reality?
• Change of price index is affected by 2 parts, i.e. the interest rate
and current price index.
π₯ =π§+ π¦−π π₯
π¦ = 1 − ππ¦ − π₯ 2
π§ = −π₯ − ππ§
Balance point: π₯ = π¦ = π§ = 0
β π=
1
0, , 0
π
if π − π − πππ ≤ 0 (β΅ ππ₯ 2 = π − π − πππ)
1
π−π−πππ 1+ππ
1
,
,β
π
π
π
β‘ π = 0, π , 0 , Q± = ±
π−π−πππ
π
if π − π − πππ ≥ 0
Stability theorem (Jacobian-->eigenvalue)
If all eigenvalues of J have strictly negative real part then the solution is stable.
If all eigenvalues of J have strictly positive real part then the solution is unstable.
If the eigenvalues of J have strictly negative real part and strictly positive real
part then the solution is a saddle.
1
β Set π = π₯, π = π¦ − π , π = π§ βΉ
1
−π
π
π=
π + π + ππ
and P = X0 , Y0 , π0 = 0,0,0
π = −ππ − π 2
π = −π − ππ
π=
β‘ Set π = π₯ −
π−π−πππ
1+ππ
,
π
=
π¦
−
,π = π§ +
π
π
π−π−πππ
βΉ
π
π
+ π + ππ +
π
π = −ππ − π 2 − 2
π−π−πππ
π
π
π−π−πππ and Q
π
π
±
= (0,0,0)
π = −π − ππ
By noticing π ≈ π½ π0 π − π0 ,
linearize this set of non-linear equations: π½ P =
β Characteristic equation: det ππΌ − π½ = π + π
1
β Eigenvalues: π1 = −π, π2 =
− π+π−π −
1 2
π+π−π
2
1
−π
π
0
1
0
−π 0
−1
0 −π
1
π
π2 + π + π − π π + 1 + ππ − π = 0
π
−4 ππ−π+1
1
, π3
− π+π−π +
1 2
π+π−π
2
π
−4 ππ−π+1
satisfing π2 + π + π −
1
Only study on balance point π = 0, π , 0
Case 1
π
1
π − π − πππ < 0 βΉ 1 + ππ − π > 0 πππ π + π − π > 0 βΉ π1 < 0, π2 < 0, π3 < 0 (only consider real
part of eigenvalues)
• Balance point is stable convergence.
Case 2
1
π − π − πππ < 0 πππ π + π − < 0 βΉ π1 < 0, π2 > 0, π3 > 0
π
• Balance point is saddle.
Case 3
π − π − πππ > 0
π1 , π2 < 0, π3 > 0 βΉbalance point P is a saddle point.
x-t diagramοΌinitial value
(x=1,y=3,z=8)
a=0.3
a=6
a=2
a=9
y-t diagram οΌinitial value x=1,y=3,z=8)
a=0.3
a=6
a=2
a=9
z-t diagram οΌinitial value x=1,y=3,z=8)
a=0.3
a=6
a=2
a=9
x-t diagram a=2
initial value (x=1, y=3, z=8)
initial value (x=1, y=0, z=8)
initial value (x=8, y=3, z=8)
initial value (x=1, y=3, z=2)
a=1
a=5
a=9
initial value x=1,y=3,z=8
time delay
• What is time delay?
• One is the time lag between the time economic decisions are made and the time the
decisions bear fruit.
•
• simplified finance model
• By adding time-delayed feedbacks to system, we can get the new system, where s1,
s2, and s3 are time delays and K1, K2, and K3 indicate the strengths of the feedbacks.
• Hopf bifurcation of the model with time delay
• we only discuss the time delay on Y (the investment demand)
• Lemma1 When c−b−abc≤0, i.e.,1+ac−bc >0, system (1) has a unique equilibrium
• P0 (0,1/b,0)
• We calculated X=sqrt(1-ab-b/c) or X=-sqrt(1-ab-b/c)
• So if c−b−abc≤0, i.e., 1-ab-b/c≤0. The system has a single equilibrium iff 1-ab-b/c=0.
• So X=0. Substitute and get P0 (0,1/b,0)
• When c−b−abc > 0, we have 2 equilibrium solutions. X=sqrt(1-ab-b/c) or X=-sqrt(1ab-b/c)
• Substitute and get P+ and P• So we have three equilibria in total.
•
• First Let x=x, y=y−1/b, z=z
• linearization of system (2) at the equilibrium (0, 0, 0) can be written as:
• We know that the equilibrium point will not be stable when both exponentials in the
solution converge to 0, which implies that both eigenvalues need to be smaller than 0.
• So we need to prove when 1+ac−bc >0 and c+a−1/b >0, λ1 λ2 are real and smaller
than 0.
• We can use lemma4 to prove.
•
• Lemma 4 The two roots λ1,2 = 1/2(−P ±sqrt(P −4Q)) of (4) have always negative real
parts when τ ≥ 0 with the condition (H1)
• Let P = c+a – 1/b , Q = 1 + ac − bc ; it’s easy to prove P2−4Q>0. So λ1,2 are real.
•
•
Code and Graph
Evolution of x and z
• Interpretation of Time History Plots
1. Stable Oscillations:
stable cyclic pattern, common
energy is conserved
2. Damped Oscillations:
loses energy over time, moving towards equilibrium
3. Unstable Oscillations:
The system is unstable and may exhibit chaotic behavior.
Purpose of the model
• By analyzing these behaviors
• fine-tune the parameters to achieve desired system dynamics
• such as ensuring stable oscillations for practical applications in
engineering or scientific systems.
Thanks a lot!