Research Article
Asset Price Dynamics in a Chartist-Fundamentalist Model with
Time Delays: A Bifurcation Analysis
Copyright © 2016 Loretti I. Dobrescu et al. This is an open access article distributed under the Creative Commons Attribution
License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly
cited.
This paper studies the dynamic behavior of asset prices using a chartist-fundamentalist model with two speculative markets. To
this effect, we employ a differential system with delays à la Dibeh (2007) to describe the price dynamics and we assume that the two
markets are coupled via diffusive coupling terms. We study two different time delay cases, namely, when both markets experience
the same time delay and when the time delay is different across markets. First, we theoretically determine that the equilibrium exists
and investigate its stability. Second, we establish the general conditions for the existence of local Hopf bifurcations and analyze their
direction and stability. The common conclusion from both the delay scenarios we consider is that coupled speculative markets with
heterogeneous agents in each, but with different price dynamics, can be synchronized through diffusive coupling. Finally, we provide
some numerical illustrations to confirm our theoretical findings.
1. Introduction sell an asset. Specifically, Dibeh [5] used the following delay-
differential equation to describe the dynamics of asset prices:
When it comes to financial markets dynamics, there is wide
consensus that the Efficient Market Hypothesis (EMH) is the 𝑝̇ (𝑡) = (1 − 𝑚) tanh (𝑝 (𝑡) − 𝑝 (𝑡 − 𝜏)) 𝑝 (𝑡)
standard theory [1]. This theory however fails to explain or (1)
− 𝑚 (𝑝 (𝑡) − V) 𝑝 (𝑡) ,
predict situations when, for instance, speculative booms are
followed by severe crashes. In other words, it cannot explain where 𝑝(𝑡) is the actual asset price, V is the fundamental
the excess volatility in these financial markets. As a result, asset price, and 𝑚 ∈ (0, 1) is the market fraction of funda-
several models have been developed in the last decade to mentalists. For chartists, the time delay 𝜏 was introduced to
describe markets fluctuations [1–12]. For instance, Dibeh [5] capture the fact that they base the slope estimation of their
considers a market with two types of participants, namely, asset price trend on adaptive expectations that consider the
fundamentalists and chartists. The first categories of agents past values of the price trend slope. Simulation results showed
follow the EMH theory and base their decisions (and hence that there may exist limit cycles for (1), which could explain
their demand formation) on the difference between the actual the persistence of deviations from the fundamental price in
asset price and the fundamental asset price. On the other speculative markets. These findings are crucial, as Bouchaud
hand, the chartists base their market participation decision and Potters [13] found that, in models considering one asset
on the price trend of an asset. Thus, they attempt to exploit (i.e., one market), feedback mechanisms are the main driver
past price information when deciding whether to purchase or of market fluctuations, booms, and crashes.
2 Discrete Dynamics in Nature and Society
Using Hassard et al. [14], Qu and Wei [1] provided a formulation represents the chartists’ belief that price growth
theoretical justification for the results in Dibeh [5]. Applying rates are bounded and so it introduces a saturation effect
the local Hopf bifurcation theory, they analytically inves- into the chartists demand function. Previous results showed
tigated the existence of periodic oscillations for (1), which that the synchronization of coupled speculative markets with
depends on both the time delay 𝜏 and the market fraction different dynamics can occur through diffusive coupling.
of fundamentalists 𝑚. Using the normal form theory and When going from an uncoupled to a coupled market model,
center manifold theorem (see also [15, 16]), they derived the the stable converging dynamics is replaced with limit cycle
sufficient conditions to determine the direction of Hopf bifur- oscillations around the fundamental prices. Thus, coupling
cation and the stability of the bifurcating periodic solutions. and contagion between financial markets can be responsible
In modern finance theory however one of the key ideas is for the transmission of fluctuations across these markets and
portfolio diversification. And once two or more risky assets appear crucial for their stability [19].
(or markets) are available to investors (beliefs on) correlation Here is the rationale. The global financial system is
in price/returns also becomes an investment decision factor. formed of a multitude of very diverse markets, located all
But could agents’ beliefs about returns correlation generate around the world, and those trade wide-ranging classes of
comovements in risky assets prices? And how can markets assets. One of their common denominators is that assets price
become interdependent? For instance, to what extent may changes often respond to the same economic information and
the price dynamics in one market be affected by changes in market news [20–22]. This dependency on the same signals
agents’ behavior and beliefs in the alternative market? leads to price variations that are often correlated. In other
In one of the first attempts to tackle these questions, words, the price time series can exhibit similar characteristics,
Chiarella et al. [17] developed a discrete time model that which implies that there is “coupling” between markets.
Moreover, Fenn et al. [19, 23] found that the strength of cor-
combines the chartists-fundamentalists setup with the clas-
relations between many different assets increases following
sical model of diversification between one risk-free and two
a credit crisis (i.e., 2007-2008 one), as financial institutions
risky assets. Their analysis showed how chartists’ beliefs and hold similar portfolios of assets [24]. This has important
behavior may cause increasingly irregular price fluctuations implications for the robustness of financial markets [25]. If
that can be transmitted from one market to another. Fur- many assets are correlated and prices fall, this can cause
thermore, extending his previous work on models involving several financial institutions to write down the value of their
one asset (or one market), Dibeh [18] studied the role assets. And these write-downs can then significantly impact
of feedback mechanism in synchronization and contagion the credit relationships between different institutions [26].
effects (i.e., high asset prices correlation) between different The aim of this paper is to provide a detailed theoretical
markets. To this effect, the author used a nonlinear chartist- analysis of the phenomenon of coupling and contagion
fundamentalist model with two markets coupled through a between markets from the bifurcation point of view. First, we
position feedback mechanism given by the price differential will analytically prove that a model involving two financial
between the asset prices. markets with correlated prices has equilibrium points that are
Drawing on the work in Dibeh [18], in this paper we locally asymptotically stable. Second, for these equilibrium
consider the following system: points, we will determine the specific values 𝜏1 and 𝜏2 (i.e.,
the time delay parameters of the two markets) for which a
Hopf bifurcation occurs. Finally, we will study the direction
𝑝̇ 1 (𝑡) = (1 − 𝑚1 ) tanh (𝑝1 (𝑡) − 𝑝1 (𝑡 − 𝜏1 )) 𝑝1 (𝑡) of the Hopf bifurcation, as well as the stability and period of
the bifurcating periodic solutions.
− 𝑚1 (𝑝1 (𝑡) − V1 ) 𝑝1 (𝑡) We choose the time delays 𝜏1 and 𝜏2 as bifurcation param-
eters as they represent the “memory” of financial markets.
+ 𝑞1 (𝑝2 (𝑡) − 𝑝1 (𝑡)) , Frank [27] discusses extensively the importance of time
(2) delays in modeling financial markets memory and their effect
𝑝̇ 2 (𝑡) = (1 − 𝑚2 ) tanh (𝑝2 (𝑡) − 𝑝2 (𝑡 − 𝜏2 )) 𝑝2 (𝑡) on the qualitative behavior of asset prices. Relatedly, in a 2001
paper, LeBaron [28] also discusses the importance of time
− 𝑚2 (𝑝2 (𝑡) − V2 ) 𝑝2 (𝑡) horizons in agent-based computational economics. Other
works that use time delays as parameters of bifurcations
+ 𝑞2 (𝑝1 (𝑡) − 𝑝2 (𝑡)) , include Hale and Lunel [29], Hassard et al. [14], Mircea et al.
[10], Qu and Wei [1], and Xu and Li [30].
The paper proceeds as follows. In Section 2 we determine
where 𝑚𝑖 ∈ (0, 1), 𝑖 = 1, 2, are the market shares of the equilibrium points of differential system (2) for V1 ≠
fundamentalists, 𝜏𝑖 ≥ 0, 𝑖 = 1, 2, are time delays, V𝑖 > 0, V2 . Section 3 investigates the local asymptotic stability of the
𝑖 = 1, 2, denote the fundamental asset prices, 𝑞𝑖 ≥ 0, 𝑖 = 1, 2, equilibrium points and establishes the existence of the Hopf
represent the coupling strength between the two markets, and bifurcation for these points. Section 4 deals with the direction
𝐼𝑡 = 𝑝1 (𝑡) − 𝑝2 (𝑡) denotes an asset’s market index. Note that and the stability of the Hopf bifurcation. In Section 5 we use
the expectations function of the chartists is nonlinear, which numerical simulations to illustrate the validity of our main
is conveyed through the hyperbolic tangent function. This results. Section 6 concludes the paper.
Discrete Dynamics in Nature and Society 3
2. Existence and Stability of Proof. The proof follows from 𝑎3 > 0 and 𝑎2 < 0, for any 𝑚𝑖 ∈
the Equilibrium Points (0, 1) and V𝑖 > 0, 𝑖 = 1, 2, and 𝑎0 (𝑞20 ) = 0, 𝑎1 (𝑞21 ) = 0.
Consider the model given by system (2). Since time delays do Proposition 2. Let
not change, the equilibrium points of this model are given by
the solutions of the system: (3𝑎1 𝑎3 − 𝑎2 2 )
𝑝= ,
3𝑎3 2
𝑚1 (𝑝1 − V1 ) 𝑝1 − 𝑞1 (𝑝2 − 𝑝1 ) = 0,
(3)
𝑚2 (𝑝2 − V2 ) 𝑝2 − 𝑞2 (𝑝1 − 𝑝2 ) = 0, (27𝑎0 𝑎3 2 − 9𝑎1 𝑎2 𝑎3 + 2𝑎2 3 ) (11)
𝑞= ,
27𝑎3 3
where V𝑖 > 0 and 𝑚𝑖 ∈ (0, 1) and 𝑞𝑖 > 0, for 𝑖 = 1, 2. The
solution to system (3) is 𝑝1 = 𝑝2 = 0. To find the solution of 𝑞2 𝑝3
system (3) with 𝑝𝑖 > 0 and 𝑞𝑖 > 0, 𝑖 = 1, 2, we proceed as Δ= + ,
4 27
follows.
where 𝑎3 , 𝑎2 , 𝑎1 , and 𝑎0 are given in (8). The equilibrium point
First, from (3) it follows that
of system (2) is 𝐸0 (𝑝10 , 𝑝20 ), where
𝑝2 1/3 1/3
𝑝1 = (𝑞 − 𝑚2 V2 + 𝑚2 𝑝2 ) , 𝑞 𝑞 𝑎
𝑞2 2 𝑝20 = (− + √Δ) + (− − √Δ) − 2 ,
(4) 2 2 (3𝑎3 )
𝑝 (12)
𝑝2 = 1 (𝑞1 − 𝑚1 V1 + 𝑚1 𝑝1 ) , 𝑝 (𝑚 𝑝 − 𝑚2 V2 + 𝑞2 )
𝑞1 𝑝10 = 20 2 20 .
𝑞2
which yields For 𝑚𝑖 ∈ (0, 1) and V𝑖 > 0, 𝑖 = 1, 2 fixed, the coordinates of 𝐸0
𝑞 are positive.
𝑝𝑖 ≥ V𝑖 + 𝑖 , 𝑖 = 1, 2. (5)
𝑚𝑖
Proof. The proof follows from Lemma 1 and Cardan’s for-
Since 𝑝𝑖 > 0, from (5) it follows that the coupling mula, which give the positive solution of a 3rd-degree
coefficients 𝑞𝑖 , 𝑖 = 1, 2, satisfy the conditions: equation.
Let 𝐸0 (𝑝10 , 𝑝20 ) be an equilibrium point, different from
𝑞𝑖 < 𝑚𝑖 V𝑖 , 𝑖 = 1, 2. (6) (0,0), and let 𝑢1 (𝑡) = 𝑝1 (𝑡) − 𝑝10 , and 𝑢2 (𝑡) = 𝑝2 (𝑡) − 𝑝20 .
System (2) then becomes
From (4) it follows that 𝑝2 satisfies the following equation:
𝑢̇ 1 (𝑡) = 𝑎11 𝑢1 (𝑡) + 𝑎12 𝑢2 (𝑡) + 𝑏11 𝑢1 (𝑡 − 𝜏1 ) + 𝐹1 ,
𝑎3 𝑝2 3 + 𝑎2 𝑝2 2 + 𝑎1 𝑝2 + 𝑎0 = 0, (7) (13)
𝑢̇ 2 (𝑡) = 𝑎21 𝑢1 (𝑡) + 𝑎22 𝑢2 (𝑡) + 𝑏22 𝑢2 (𝑡 − 𝜏2 ) + 𝐹2 ,
where where
2 𝑎11 = (1 − 3𝑚1 ) 𝑝10 + 𝑚1 V1 − 𝑞1 ,
𝑎1 = 𝑚1 (𝑚2 V2 − 𝑞2 ) − 𝑞2 𝑚2 (𝑚1 V1 − 𝑞1 ) ,
with 𝑙𝑖 , 𝑘𝑖 , 𝑖 = 1, 2, being given by 0 are relevant. The other two cases (that involve 𝜏2 = 0)
are however needed methodologically, as they allow us to
𝑙1 = (1 − 2𝑚1 ) 𝑝10 , derive some of the results that we need when analyzing the
economically relevant cases.
𝑙2 = − (1 − 2𝑚1 ) 𝑝10 ,
𝑝10 Case 1. If 𝜏1 = 𝜏2 = 0, then
𝑙3 = − (1 − 𝑚1 ) ,
3 𝑙 (𝜆, 0, 0) = 𝜆2 − (𝛽1 + 𝛽3 + 𝛽5 ) 𝜆 + 𝛽2 + 𝛽4 + 𝛽6 + 𝛽7 . (19)
𝑙4 = (1 − 𝑚1 ) 𝑝10 , Using (14) and (18), the equation 𝑙(𝜆, 0, 0) = 0 becomes
Using cos(𝜔𝜏)2 + sin(𝜔𝜏)2 = 1, from (26), it follows that Assuming that the following conditions hold:
𝜔8 + 𝛾6 𝜔6 + 𝛾4 𝜔4 + 𝛾2 𝜔2 + 𝛾0 = 0, (27) 𝛾0 < 0,
2 2 2 2
[arccos (((𝜔10 − 𝛽2 − 𝛽6 ) (𝛽2 + 𝛽7 ) − 𝛽3 (𝛽1 + 𝛽5 ) 𝜔10 ) / ((𝛽1 + 𝛽7 ) + 𝛽3 𝜔10 )) + 2𝑗𝜋]
𝜏1𝑗 = , (42)
𝜔10
2 2
with 𝑗 = 0, 1, 2, . . .. If the conditions hold, then (39) has two positive roots 𝜔10+ and 𝜔10− . Substi-
2
tuting 𝜔10± into (38), we obtain
𝛿2 < 0,
(43)
𝛿0 > 0
2 2 2 2
[arccos (((𝜔10± − 𝛽2 − 𝛽0 ) (𝛽2 + 𝛽7 ) − 𝛽3 (𝛽1 + 𝛽5 ) 𝜔10± ) / ((𝛽1 + 𝛽7 ) + 𝛽3 𝜔10± )) + 2𝑘𝜋]
𝜏1𝑘± = , (44)
𝜔10±
We assume that there exists 𝜔0 > 𝑝 such that 𝐾21 (𝜔0 ) = 0. 𝜏1 = 𝜏10 + 𝜇, where 𝜏10 is given by (42) and 𝜇 ∈ R. Differential
For 𝜔0 , there exists a sequence {𝜏2𝑗 , 𝑗 = 1, 2, . . .} such that system (2) can be written as a FDE in 𝐶 = 𝐶([−1, 0], R2 ) as
(50) holds. follows:
Let 𝜏21 = max{𝜏2𝑖 , 𝑖 = 1, . . .}. When 𝜏2 = 𝜏21 , (17) has a
pair of purely imaginary roots ±𝑖𝜔21 , for 𝜏1 ∈ [0, 𝜏10 ). 𝑢̇ (𝑡) = 𝐿 𝜇 (𝑢𝑡 ) + 𝐹 (𝜇, 𝑢𝑡 ) , (56)
In the following we assume that
where 𝑢(𝑡) = (𝑢1 (𝑡), 𝑢2 (𝑡))𝑇 ∈ 𝐶 and 𝑢𝑡 (𝜙) = 𝑢(𝑡 + 𝜃) =
𝑑 Re (𝜆 (𝜏)) −1 (𝑢1 (𝑡 + 𝜃), 𝑢2 (𝑡 + 𝜃))𝑇 ∈ 𝐶. The terms 𝐿 𝜇 : 𝐶 → R and
( ) ≠ 0. (53) 𝐹 : R × 𝐶 → R are, respectively, given by
𝑑𝜏2 𝜆=𝑖𝜔21
𝜏2
Thus, by the general Hopf bifurcation theorem for FDEs in 𝐿 𝜇 𝜙 = 𝐴𝜙 (0) + 𝐵1 𝜙 (−1) + 𝐵2 𝜙 (− ),
𝜏10 (57)
Hale and Lunel [29], we have the following result on stability
and Hopf bifurcation for system (2). 𝑇
𝐹 (𝜇, 𝜙) = (𝐹11 (𝜙) , 𝐹12 (𝜙)) ,
Proposition 5. Assume that the above condition, where 𝜙(𝜃) = (𝜙1 (𝜃), 𝜙2 (𝜃))𝑇 ∈ 𝐶 and
(𝑑 Re(𝜆(𝜏))/𝑑𝜏2 )−1
𝜆=𝑖𝜔21 ≠ 0, is satisfied and that
𝜏1 ∈ [0, 𝜏10 ). Then, the equilibrium point 𝐸0 (𝑝10 , 𝑝20 ) is 𝑎11 𝑎12
𝐴=( ),
locally asymptotically stable when 𝜏2 ∈ [0, 𝜏21 ). Moreover, 𝑎21 𝑎22
when 𝜏2 = 𝜏21 , system (2) undergoes Hopf bifurcation at
𝐸0 (𝑝10 , 𝑝20 ). 𝑏11 0
𝐵1 = ( ),
0 0
If 𝜔21 is a positive root of (50), then 𝜏2𝑗 is given by
0 0 (58)
1 𝐻 𝐵2 = ( ),
𝜏2𝑗 = [arccos ( 21 ) + 2𝑗𝜋] , 𝑗 = 0, 1, 2, . . . , (54) 0 𝑏22
𝜔21 𝐿 21
where 𝐹11 (𝜙) = (𝜏10 + 𝜇) 𝐹1 (𝜙1 (0) , 𝜙1 (−1)) ,
𝐻21 𝜏2
𝐹12 (𝜙) = (𝜏10 + 𝜇) 𝐹2 (𝜙2 (0) , 𝜙2 (− )) ,
𝜏10
2
= (𝜔21 − 𝛽2 − 𝛽1 cos (𝜔21 𝜏1 ) + 𝛽3 𝜔21 sin (𝜔21 𝜏1 ))
with 𝐹1 and 𝐹2 being given by (13).
⋅ (𝛽6 + 𝛽7 cos (𝜔21 𝜏1 )) From Section 3 we know that if 𝜇 = 0, then differential
system (2) undergoes a Hopf bifurcation at 𝐸0 (𝑝10 , 𝑝20 ) and
− (𝜔21 𝛽1 + 𝛽3 𝜔21 cos (𝜔21 𝜏1 ) + 𝛽1 sin (𝜔21 𝜏1 )) (55) the characteristic equation associated with system (2) has
a pair of simple imaginary roots ±𝑖𝜔10 𝜏10 . By the Reisz
⋅ (𝛽2 𝜔21 + 𝛽7 sin (𝜔21 𝜏1 )) , representation theorem, there is a matrix function with
bounded variation components 𝜂(𝜃, 𝜇), 𝜃 ∈ [−1, 0] such that
2
𝐿 21 = (𝛽6 + 𝛽7 cos (𝜔21 𝜏1 )) 0
2 𝐿 𝜇 𝜙 = ∫ 𝑑𝜂 (𝜃, 𝜇) 𝜙 (𝜃) , for 𝜙 ∈ 𝐶. (59)
+ (𝛽3 𝜔21 + 𝛽7 sin (𝜔21 𝜏1 )) . −1
We can choose
4. The Direction and the Stability of 𝐴 1 + 𝐵11 + 𝐵12 , 𝜃=0
the Hopf Bifurcation {
{
{
{
{
{ 𝜏2
{
{𝐵11 + 𝐵12 , 𝜃 ∈ [− , 0)
In the previous section, we have obtained the condition for 𝜂 (𝜃, 𝜇) = { 𝜏10 (60)
the Hopf bifurcation to occur when 𝜏1 = 𝜏10 , where 𝜏10 is {
{ 𝜏
{
{ 𝐵11 , 𝜃 ∈ (−1, − 2 )
given by (42). Using the techniques from normal form and {
{ 𝜏10
{
center manifold theory introduced by Hassard et al. [14], we {0, 𝜃 = −1.
now derive the explicit formulae determining the direction,
stability, and periodicity of these solutions bifurcating from For 𝜙 ∈ 𝐶([−1, 0], R3 ), we define
the positive equilibrium 𝐸0 (𝑝10 , 𝑝20 ) at the critical value 𝜏1 =
𝜏10 . To this effect, we will use the first Lyapunov coefficient. 𝑑𝜙 (𝜃)
{
Throughout this section, we assume that differential { 𝑑𝜃 ,
{ 𝜃 ∈ [−1, 0)
𝐴 (𝜇) 𝜙 (𝜃) = { 0
system (2) undergoes a Hopf bifurcation at the positive {
{∫ 𝑑𝜂 (𝑠, 𝜇) 𝜙 (𝑠) , 𝜃 = 0,
equilibrium 𝐸0 (𝑝10 , 𝑝20 ) for 𝜏1 = 𝜏10 and that ±𝑖𝜔10 are { −1 (61)
the corresponding purely imaginary roots of the character-
istic equation at 𝐸0 (𝑝10 , 𝑝20 ). Without loss of generality, we {0, 𝜃 ∈ [−1, 0)
assume that 𝜏2 < 𝜏10 , where 𝜏2 ∈ (0, 𝜏21 ) and 𝜏21 is given 𝑅 (𝜇2 ) 𝜙 = {
𝐹 (𝜇, 𝜙) , 𝜃 = 0.
in (54). For convenience, let 𝑢𝑖 (𝑡) = 𝑢𝑖 (𝜏1 𝑡), 𝑖 = 1, 2, and {
8 Discrete Dynamics in Nature and Society
Then, (57) is equivalent to the abstract differential equation: We now turn to studying the stability of the bifurcating
periodic solution. As in Xu and Li [30], the bifurcating
𝑢̇ 𝑡 (𝑡) = 𝐴 (𝜇) 𝑢𝑡 + 𝑅 (𝜇) 𝑢𝑡 , (62) periodic solution 𝑍(𝑡, 𝜇(𝜀)) has amplitude 𝑂(𝜀) and nonzero
Floquet exponent 𝛽(𝜀), with 𝛽(0) = 0. Under these assump-
where 𝑢𝑡 = 𝑢(𝑡 + 𝜃), for 𝜃 ∈ [−1, 0]. tions, 𝜇 and 𝛽 are given by
For 𝜙 ∈ 𝐶([−1, 0], (R2 )∗ ), we define
−𝑑𝜓 (𝑠) 𝜇 = 𝜇2 𝜀2 + 𝜇4 𝜀4 + ⋅ ⋅ ⋅ ,
{
∗ { 𝑑𝑠 ,
{ 𝑠 ∈ (0, 1] (71)
𝐴 𝜓 (𝑠) = { 0 (63) 𝛽 = 𝛽2 𝜀2 + 𝛽4 𝜀4 + ⋅ ⋅ ⋅ .
{
{∫ 𝑑𝜂𝑇 (𝑡, 0) 𝜓 (𝑠) , 𝑠 = 0.
{ −1 The sign of 𝜇2 indicates the direction of the bifurcation, while
2
For 𝜙 ∈ 𝐶([−1, 0], R ) and 𝜓 ∈ 𝐶([0, 1], (R ) ), we define the 2 ∗ 𝛽2 determines the stability of 𝑍(𝑡, 𝜇(𝜀)), with 𝑍(𝑡, 𝜇(𝜀)) being
bilinear form stable if 𝛽2 < 0 and unstable if 𝛽2 > 0.
In what follows, we will show how to derive these
⟨𝜓, 𝜙⟩ = 𝜓 (0) 𝜙 (0) coefficients (i.e., 𝜇2 and 𝛽2 ). To this effect, we first construct
the coordinates needed to describe the center manifold Ω0
0 𝜃 (64)
near 𝜇 = 0, which is a local invariant, attracting a two-
−∫ ∫ 𝜓𝑇 (𝜉 − 𝜃) 𝑑𝜂 (𝜃) 𝜙 (𝜉) 𝑑𝜉,
−1 𝜉=0 dimensional manifold [30].
Let 𝑢𝑡 be the solution of (63) when 𝜇 = 0, and define
where 𝜂(𝜃) = 𝜂(𝜃, 0), 𝐴 = 𝐴(0), and 𝐴∗ are adjoint operators.
Following from Section 3 discussion, we also know that 𝑧 (𝑡) = ⟨𝑞∗ , 𝑢𝑡 ⟩ ,
±𝑖𝜔10 𝜏10 are eigenvalues of 𝐴(0), but they are also eigenvalues (72)
of 𝐴∗ . 𝑊 (𝑡, 𝜃) = 𝑢𝑡 (𝑡, 𝜃) − 2 Re (𝑧 (𝑡) , 𝑞 (𝜃)) .
Proposition 6. (i) If the eigenvector 𝑞 of 𝐴(𝜇) is given by On the center manifold Ω0 , we have 𝑊(𝑡, 𝜃) = 𝑊(𝑧(𝑡),
𝑇 𝑖𝜔10 𝜏10 𝜃 𝑧(𝑡), 𝜃), where
𝑞 (𝜃) = (1, 𝛼) 𝑒 , 𝜃 ∈ (−1, 0] , then (65)
p1 (t), p2 (t)
(78) 16
𝑇2 = − ,
𝜔10 𝜏10
14
𝛽2 = 2 Re (𝐶1 (0)) .
12
Using these results, we establish the following theorem on
the system trajectories properties. 10
4
5. Numerical Illustrations
3.5
To illustrate our results, we employ several numerical sim-
ulations. To this purpose, we set 𝑚𝑖 ∈ (0, 1) and V𝑖 > 0
3
(with 𝑖 = 1, 2) and choose 𝑞1 , 𝑞2 according to Lemma 1 and 0 50 100 150 200 250 300 350
Proposition 2. Specifically, let 𝑚1 = 0.2, 𝑚2 = 0.3, V1 = Time t
20, V2 = 10, and 𝑞1 = 2.5 < 𝑚1 V1 = 4. Using Lemma 1
and Proposition 2, we set 𝑞2 = 1.525. From Proposition 2, we Figure 2: (𝑡, 𝐼(𝑡)), for 𝜏 < 𝜏0 .
then get 𝑎3 = 0.18, 𝑎2 = −0.77, 𝑎1 = −0.251, and 𝑎0 = −2.44.
Using (9), we have that 𝑝10 = 16.53 and 𝑝20 = 11.95. With
this set of parameters, we obtain the following results: 20
(1) If 𝜏1 = 𝜏2 = 0, from (20) it follows that 𝜆 1 = −3.4302 19
and 𝜆 2 = −7.378. The system is therefore locally
asymptotically stable. 18
16
For 𝜏 < 𝜏0 , the system is asymptotically stable; for
𝜏 = 𝜏0 , it has a Hopf bifurcation; and for 𝜏 > 𝜏0 , it 15
is unstable.
14
The orbits (𝑡, 𝑝1 (𝑡)) (in black) and (𝑡, 𝑝2 (𝑡)) (in blue) are 13
given in Figures 1, 3, and 5 for 𝜏 < 𝜏0 , 𝜏 = 𝜏0 , and 𝜏 > 𝜏0 ,
respectively. Similarly, Figures 2, 4, and 6 plot the market 12
index (𝑡, 𝐼(𝑡)), where 𝐼(𝑡) = 𝑝1 (𝑡) − 𝑝2 (𝑡), for each of the three 11
cases. 0 20 40 60 80 100 120 140 160 180
Time t
(3) If 𝜏1 < 𝜏10 , 𝜏2 = 0; from (39) it follows that 𝜔10 =
12.265. From (42) it follows that 𝜏10 = 0.57185. For Figure 3: (𝑡, 𝑝1 (𝑡)) and (𝑡, 𝑝2 (𝑡)), for 𝜏 = 𝜏0 .
10 Discrete Dynamics in Nature and Society
7 16.5
6.5
16
6
5.5 15.5
p1 (t), p2 (t)
5
I(t)
15
4.5
4 14.5
3.5
14
3
2.5 13.5
0 20 40 60 80 100 120 140 160 180 0 50 100 150 200 250
Time t Time t
Figure 4: (𝑡, 𝐼(𝑡)), for 𝜏 = 𝜏0 . Figure 7: (𝑡, 𝑝1 (𝑡)) and (𝑡, 𝑝2 (𝑡)), for 𝜏 < 𝜏0 .
0.8
20
0.6
19
18 0.4
17 0.2
I(t)
p1 (t), p2 (t)
16
0
15
−0.2
14
13 −0.4
12
−0.6
0 50 100 150 200 250
11
0 50 100 150 200 250 300 Time t
Time t Figure 8: (𝑡, 𝐼(𝑡)), for 𝜏 < 𝜏0 .
Figure 5: (𝑡, 𝑝1 (𝑡)) and (𝑡, 𝑝2 (𝑡)), for 𝜏 > 𝜏0 .
6. Conclusions Appendices
This paper considers a deterministic dynamic model with A. Data of Lemma A.1
delay for two financial markets that trade one asset. The deter-
ministic system is analyzed both analytically and numerically. Lemma A.1 (see [31, Corollary 2.4, page 867]). For the
We study two different cases of time delay, namely, when both transcendental equation
markets experience the same time delay and when the time
𝑛
delay is different across markets. The common conclusion
∑𝑃 (𝜆, 𝑒−𝜆𝜏1 , . . . , 𝑒−𝜆𝜏𝑛 )
from these two scenarios is that coupled speculative markets 𝑖=0
with chartists and fundamentalists in each, but with different (A.1)
𝑛
dynamics, can be synchronized through diffusive coupling. 𝑛 𝑖 −𝜆𝜏1 −𝜆𝜏𝑛
The stable converging dynamics is replaced with limit cycle = 𝜆 + ∑𝑃 (𝜆, 𝑒 ,...,𝑒 ) = 0,
𝑖=0
oscillations around the fundamental price. This qualitative
change in the dynamic behavior shows that coupling and where
contagion between different markets can indeed lead to
transmission of fluctuations across financial markets. Since 𝑃𝑖 (𝜆, 𝑒−𝜆𝜏1 , . . . , 𝑒−𝜆𝜏𝑛 )
the delay can lead to instability, the next natural step would
be to consider the stochastic approach. We leave this issue for = [𝛽1(𝑖) 𝜆𝑛−1 + ⋅ ⋅ ⋅ + 𝛽𝑛−1
(𝑖)
𝜆 + 𝛽𝑛(𝑖) ] 𝑒−𝜆𝜏𝑖 , (A.2)
future research.
In terms of contribution, our work ultimately provides a 𝑖 = 0, 1, . . . , 𝑛,
real-world application of bifurcation concepts to a situation
widely observed in the financial world. To this effect, we pro- as (𝜏1 , . . . , 𝜏𝑛 ) vary, the sum of the orders of the zeros of
vide a rigorous analysis of the dynamic properties of a model 𝑃(𝜆, 𝑒−𝜆𝜏1 , . . . , 𝑒−𝜆𝜏𝑛 ) in the open right half plane can change
that can successfully explain stylized facts related to financial only if a zero appears on or crosses the imaginary axis.
market behavior. Once established, the same methodology
could be easily implemented in economics, demography, and Proof of Proposition 6. (i) Let 𝑞(𝜃) be given by (66). It follows
social sciences in general, allowing further literature advances from above that
in these fields. For instance, this application fits well and can
be quickly integrated in the structural work done in these 𝐴𝑞 (0) = 𝑖𝜔10 𝜏10 𝑞 (0) , (A.3)
fields.
implying
𝜏2
From (74), (76), and (B.4), we obtain 𝑢2𝑡 (− ) = 𝛼𝑧𝑒−𝑖𝜔10 𝜏2 /𝜏10
𝜏10
𝑊̇ = (𝑊20 𝑧 + 𝑊11 𝑧 + ⋅ ⋅ ⋅) (𝑖𝜔10 𝜏10 𝑧 + 𝑔) 𝜏2 𝑧2
(B.5) + 𝛼𝑧𝑒𝑖𝜔10 𝜏2 /𝜏10 𝑊20
(2)
(− )
+ (𝑊11 𝑧 + 𝑊02 𝑧 + ⋅ ⋅ ⋅) (−𝑖𝜔10 𝜏10 𝑧 + 𝑔) , 𝜏10 2
(2) 𝜏2
+ 𝑊11 (− ) 𝑧𝑧
and substituting (74), (B.3), and (B.2) yields 𝜏10
(2) 𝜏2 𝑧2
𝑧2 + 𝑊02 (− ) + ⋅⋅⋅ .
𝑊̇ = (𝐴𝑊20 + ℎ20 ) + (𝐴𝑊11 + ℎ11 ) 𝑧𝑧 𝜏10 2
2
(B.6) (B.10)
𝑧2
+ (𝐴𝑊02 + ℎ02 ) + ⋅⋅⋅ .
2
Proposition B.1. The coefficients 𝑔20 , 𝑔11 , 𝑔02 , and 𝑔21 are
given by
Comparing the coefficients of (B.5) and (B.6), the following
equalities hold:
2𝜏10
(𝐴 − 2𝑖𝜔10 𝜏10 𝐼) 𝑊20 (𝑡) = −ℎ20 (𝜃) , (B.7) 𝑔20 = (𝑙1 + 𝑙2 𝑒−𝑖𝜔10 𝜏10 + 𝛼2 𝛼∗ (𝑘1
𝜌
𝐴𝑊11 (𝑡) = −ℎ11 (𝜃) , (B.8)
+ 𝑘2 𝑒−𝑖𝜔10 𝜏2 /𝜏10 )) ,
Proposition B.2. The vectors, 𝑊11 (𝜃) and 𝑊20 (𝜃), are given
by 𝑖𝑔11
𝑊11 (𝜃) = − 𝑞 (0) 𝑒𝑖𝜔10 𝜏10 𝜃
𝜔10 𝜏10
𝑖𝑔20 𝑔11
𝑊20 (𝜃) = − 𝑞 (0) 𝑒𝑖𝜔10 𝜏10 𝜃 − 𝑞 (0) 𝑒−𝑖𝜔10 𝜏10 𝜃 + 𝐸1 ,
𝜔10 𝜏10 𝑖𝜔10 𝜏10
(B.12)
𝑖𝑔02
− 𝑞 (0) 𝑒−𝑖𝜔10 𝜏10 𝜃 + 𝐸2 𝑒2𝑖𝜔10 𝜏10 𝜃 ,
3𝜔10 𝜏10 where
−1
2𝑖𝜔10 𝜏10 − 𝑎11 − 𝑏11 𝑒−2𝑖𝜔10 𝜏10 −𝑎12 2𝑙1 + 2𝑙2 𝑒−𝑖𝜔10 𝜏10
𝐸2 = ( ) ( 2 ),
−𝑎21 2𝑖𝜔10 𝜏10 − 𝑎22 − 𝑏22 𝑒−2𝑖𝜔10 𝜏2 /𝜏10 2𝛼 (𝑘1 + 𝑘2 𝑒−𝑖𝜔10 𝜏2 /𝜏10 )
(B.13)
−1
−𝑎11 − 𝑏11 −𝑎12 𝑙1 + 𝑙2 (𝑒𝑖𝜔10 𝜏10 + 𝑒−𝑖𝜔10 𝜏10 )
𝐸1 = ( ) ( ).
−𝑎21 −𝑎22 − 𝑏22 𝛼𝛼 (𝑘1 + 𝑘2 (𝑒𝑖𝜔10 𝜏2 /𝜏10 + 𝑒−𝑖𝜔10 𝜏2 /𝜏10 ))
Proof. First, we use (B.2) to obtain From (B.15) and the definition of 𝐹(𝜇, 𝜙), we have that
Comparing the coefficients with (B.3), for 𝜃 ∈ (−1, 0], we = −𝑔11 𝑞 (𝜃) − 𝑔11 𝑞 (𝜃)
obtain
𝑙1 + 𝑙2 (𝑒𝑖𝜔10 𝜏10 + 𝑒−𝑖𝜔10 𝜏10 )
ℎ20 (𝜃) = −𝑔20 𝑞 (𝜃) − 𝑔20 𝑞 (𝜃) , + 𝜏0 ( ).
(B.15) 𝛼𝛼 (𝑘1 + 𝑘2 (𝑒𝑖𝜔10 𝜏2 /𝜏10 + 𝑒−𝑖𝜔10 𝜏2 /𝜏10 ))
ℎ11 (𝜃) = −𝑔11 𝑞 (𝜃) − 𝑔11 𝑞 (𝜃) .
Substituting the above equalities into (B.7) and (B.8), respec-
tively, and with 𝐴 = 𝐴(0) being given by (62), we get that Substituting (B.12) and (B.19) into (B.18) and using the fact
that
𝑊̇ 20 (𝜃) = 2𝑖𝜔10 𝜏10 𝑊20 (𝜃) + 𝑔20 𝑞 (𝜃) + 𝑔20 𝑞 (𝜃) ,
(B.16)
𝑊̇ 11 (𝜃) = 𝑔11 𝑞 (𝜃) + 𝑔11 𝑞 (𝜃) . 0
(𝑖𝜔10 𝜏10 𝐼 − ∫ 𝑒𝑖𝜔10 𝜏10 𝜃 𝑑𝜂 (𝜃)) 𝑞 (0) = 0,
−1
Solving (B.16) leads to (B.12). To obtain the appropriate 𝐸1 (B.20)
and 𝐸2 , we proceed as follows. 0
−𝑖𝜔10 𝜏10 𝜃
From (B.7) and (B.8), it follows that (−𝑖𝜔10 𝜏10 𝐼 − ∫ 𝑒 𝑑𝜂 (𝜃)) 𝑞 (0) = 0,
−1
which leads to
2𝑖𝜔10 𝜏10 − 𝑎11 − 𝑏11 𝑒−2𝑖𝜔10 𝜏10 −𝑎12 2𝑙1 + 2𝑙2 𝑒−𝑖𝜔10 𝜏10
( ) 𝐸2 = ( ), (B.22)
−𝑎21 2𝑖𝜔10 𝜏10 − 𝑎22 − 𝑏22 𝑒−2𝑖𝜔10 𝜏2 /𝜏10 2𝛼2 (𝑘1 + 𝑘2 𝑒−𝑖𝜔10 𝜏2 /𝜏10 )
and so
−1
2𝑖𝜔10 𝜏10 − 𝑎11 − 𝑏11 𝑒−2𝑖𝜔10 𝜏10 −𝑎12 2𝑙1 + 2𝑙2 𝑒−𝑖𝜔10 𝜏10
𝐸2 = ( ) ( ). (B.23)
−𝑎21 2𝑖𝜔10 𝜏10 − 𝑎22 − 𝑏22 𝑒−2𝑖𝜔10 𝜏2 /𝜏10 2𝛼2 (𝑘1 + 𝑘2 𝑒−𝑖𝜔10 𝜏2 /𝜏10 )
Similarly, substituting (B.12) and (B.19) into (B.17), we [6] W. A. Brock, C. H. Hommes, and F. O. O. Wagener, “Evolu-
have that tionary dynamics in markets with many trader types,” Journal
−1
of Mathematical Economics, vol. 41, no. 1-2, pp. 7–42, 2005.
−𝑎11 − 𝑏11 −𝑎12
𝐸1 = ( ) [7] C. H. Hommes, “Financial markets as nonlinear adaptive
−𝑎21 −𝑎22 − 𝑏22 evolutionary systems,” Quantitative Finance, vol. 1, no. 1, pp.
(B.24) 149–167, 2001.
𝑙1 + 𝑙2 (𝑒𝑖𝜔10 𝜏10 + 𝑒−𝑖𝜔10 𝜏10 ) [8] C. H. Hommes, “Heterogeneous agent models in economics
⋅( 𝑖𝜔10 𝜏2 /𝜏10 −𝑖𝜔10 𝜏2 /𝜏10
). and finance,” in Handbook of Computational Economics, vol. 2,
𝛼𝛼 (𝑘1 + 𝑘2 (𝑒 +𝑒 ))
no. 23, pp. 1109–1186, 2006.
[9] T. Lux, “The socio-economic dynamics of speculative markets:
We can therefore derive 𝑊20 (𝜃) and 𝑊11 (𝜃), with 𝑔𝑖𝑗 in
interacting agents, chaos, and the fat tails of return distribu-
(B.10) being determined by the parameters and the time tions,” Journal of Economic Behavior and Organization, vol. 33,
delays in system (2). no. 2, pp. 143–165, 1998.
[10] G. Mircea, M. Neamtu, and D. Opris, Uncertain, Stochastic and
Conflict of Interests Fractional Dynamic Systems with Delay. Applications, Lambert
Academic Publishing, Saarbrücken, Germany, 2011.
The authors declare that there is no conflict of interests
[11] M. Neamtu, G. Mircea, and D. Opris, “Deterministic and
regarding the publication of this paper. stochastic model for the analysis of the asset price,” in Proceed-
ings of the 13th International Conference on Mathematics and
Acknowledgment Computers in Business & Economics, pp. 165–170, 2012.
[12] R. Sethi, “Endogenous regime switching in speculative mar-
The authors are especially grateful to D. Opris, the editor, for kets,” Structural Change and Economic Dynamics, vol. 7, no. 1,
helpful comments and suggestions. pp. 99–118, 1996.
[13] J. P. Bouchaud and M. Potters, Theory of Financial Risk and
References Derivative Pricing: From Statistical Physics to Risk Management,
Cambridge University Press, Cambridge, UK, 2nd edition,
[1] Y. Qu and J. Wei, “Global Hopf bifurcation analysis for a time- 2003.
delayed model of asset prices,” Discrete Dynamics in Nature and
[14] B. D. Hassard, N. D. Kazarinoff, and Y. H. Wan, Theory and
Society, vol. 2010, Article ID 432821, 17 pages, 2010.
Applications of Hopf Bifurcation, vol. 41 of London Mathematical
[2] C. Chiarella, R. Dieci, and L. Gardini, “Speculative behaviour Society Lecture Note, Cambridge University Press, Cambridge,
and complex asset price dynamics: a global analysis,” Journal of UK, 1981.
Economic Behavior and Organization, vol. 49, no. 2, pp. 173–197,
2002. [15] E. Karaoglu and H. Merdan, “Hopf bifurcations of a ratio-
[3] C. Chiarella, R. Dieci, and X. He, “Heterogeneity, market mech- dependent predator-prey model involving two discrete matura-
anisms and asset price dynamics,” in Handbook of Financial tion time delays,” Chaos, Solitons & Fractals, vol. 68, pp. 159–168,
Markets: Dynamics and Evolution, vol. 5, pp. 277–344, Elsevier, 2014.
Philadelphia, Pa, USA, 2009. [16] L.-Y. Zhang, “Hopf bifurcation analysis in a Monod-Haldane
[4] C. Chiarella and X.-Z. He, “Heterogeneous beliefs, risk, and predator-prey model with delays and diffusion,” Applied Math-
learning in a simple asset-pricing model with a market maker,” ematical Modelling, vol. 39, no. 3-4, pp. 1369–1382, 2015.
Macroeconomic Dynamics, vol. 7, no. 4, pp. 503–536, 2003. [17] C. Chiarella, R. Dieci, and L. Gardini, “The dynamic interac-
[5] G. Dibeh, “Speculative dynamics in a time-delay model of asset tion of speculation and diversification,” Applied Mathematical
prices,” Physica A, vol. 355, no. 1, pp. 199–208, 2005. Finance, vol. 12, no. 1, pp. 17–52, 2005.
Discrete Dynamics in Nature and Society 15
International
Journal of Journal of
Mathematics and
Mathematical
Discrete Mathematics
Sciences