You are on page 1of 4

A COBWEB MODEL WITH RANDOM EXPECTATIONS

Serena Brianzoni, Università degli studi di Macerata


Cristiana Mammana, Università degli studi di Macerata
Elisabetta Michetti, Università degli studi di Macerata
Francesco Zirilli, Università di Roma ‘La Sapienza’

EXTENDED ABSTRACT

1. Introduction

The cobweb model is a dynamical system that describes price fluctuations as a result
of the interaction between demand function depending on current price and supply function
depending on expected price.
A classic definition of the cobweb model is the one given by Ezekiel (1938) who
proposed a linear model with deterministic static expectation. The least convincing elements
of the initial formulation is the linearity of the functions describing the market and its simple
expectations. For these reasons several efforts have been made over time to improve the
original model. In a number of works nonlinearities have been introduced in the cobweb
model (see Holmes and Manning (1988)) while other authors considered different kinds of
price expectations (see, among others, Nerlove (1958), Chiarella (1988), Hommes (1994),
Gallas and Nusse (1996)). More recently in Mammana and Michetti (2003, 2004) an infinite
memory learning mechanism has been introduced in the nonlinear cobweb model.
In this work we consider a stochastic nonlinear cobweb model that generalizes the
model of Jensen and Urban (1984) assuming that the representative entrepreneur chooses
between two different predictors in order to formulate his expectations:
• backward predictor: the expectation of future price is the arithmetical mean of past
observations with decreasing weights, according to a geometrical progression of ρ region;1
• forward predictor: the formation mechanism of this expectation takes into account the
market equilibrium price while considering that the current price will converge to it only in
the long run.
The representative entrepreneur chooses the backward predictor with probability q
( 0 < q < 1 ) and the forward predictor with probability (1 − q ) .
The time evolution of the price is described in the new model by a stochastic
dynamical system.2 That is since for simplicity we consider only discrete time dynamical
systems, the price solution of the previous model is a discrete time stochastic process. In
particular at any given time the price is a random variable.
With this work we begin the study of these models that can be called stochastic
cobweb models. We start from an empirical study of the model described in section 2 by
doing the statistics of a sample of trajectories of the model numerically generated. In
particular we obtain the probability density function of the random variable price as a function
of time. We study these densities in the limits: time goes to infinity, q goes to zero, q goes to
one. Some interesting observations are made.
We plan to continue the present study giving a mathematical proof of the observations
cited previously and more in general carrying out a mathematical analysis of the stochastic
cobweb models.
1
See Mammana and Michetti 2004.
2
Recent works in this direction are those by Evans and Honkapohja (1998), Branch and McGough (2005) and
Brock and Hommes (1997).
2. The model

In this study we consider a cobweb-type model with linear demand and a backward-
bending supply curve (concave parabola) as in Jensen and Urban (1984). A supply curve of
this type is economically justified, for instance, by the presence of external economies, if the
advantage that businesses do not gain from their individual expansion, but rather from the
expansion of industry as a whole.3
According to the previous considerations our model is given by:
⎪ Dt +1 = D ( Pt +1 ) = a + bPt +1




⎪⎪
S =S ( Pte )=c+dPte +e( Pte )2
⎨ t +1 (1)





⎩⎪
Dt +1=St +1
D , S are respectively the amount demanded by consumers and the amount supplied by a
representative entrepreneur, with b < 0 , d > 0 and e < 0 . In our model Pt , Pt e are
respectively the price of good at time t and the expectation at time t of the price at time t + 1 .
The market clearing equation able to obtain a quadratic and convex relationship
Pt +1 = D −1 D S ( Pt e ) (2)
between price at time t + 1 and the expected price at time t . Function (2) is topologically
conjugated to the well-known logistic map4 given by
Pt +1 = f µ ( Pt e ) = µ Pt e (1 − Pt e ), µ > 1. (3)
The composit map f µ is a function of expectations so that in what follows we consider two
different formulations for the expectation formation mechanism.

1. Backward-looking expectation
Firstly we consider the backward-looking component extracts the future market price
value expected from those observed in the past through an infinite memory process.5 The use
of this type of learning mechanism is rather ’natural’ if rationality is absent and agent has to
use all the informations available that is the historical values of prices. In our model Pt e is the
arithmetical mean of the previous states taken on by the real variable, measured with
exponentially decreasing weights as the terms of a converging geometric series of ρ , region,
called memory rate:
t
ρ t −k t
1− ρt
Pt e = ∑ Pk , ρ ∈ (0, 1), Wt = ∑ ρ t −k = . (4)
k =0 Wt k =0 1− ρ
Notice that as ρ → 0 static expectations are considered while as ρ → 1 formula (4) is the
arithmetical mean of past prices. This kind of expectations have been studied in Michetti
(2000) while in Mammana and Michetti (2004) the model herewith studied have been
considered in a deterministic contest.6
3
See Sraffa (1926).
4
The logistic map can be obtained from (2) by a linear tranformation in the variable.
5
This iterative scheme is known as Mann iteration, see Mann (1953).
6
Equation (3) with backward expectation as in (4) has an equivalent autonomous limit form describing the
e
expectation prices dynamics given by Pt +1 = − µ (1 − ρ )( Pt e )2 + [ µ (1 − ρ ) + ρ ]Pt e that can be used to study

2
2. Forward-looking expectation
Secondly we consider a forward looking expectation creation mechanism. This
component aims at providing an answer to the criticism put forward by economists on the
lack of rationality in the expectations introduced in price-quantity models. We assume that
representative supplier knows the market equilibrium price, namely P ƒ , but at the same time
he knows that the process leading to the balance is not instantaneous (therefore we do not
assume totally rational expectations as in Mammana and Michetti (2004)). This assumption is
consistent with the conclusion reached in many dynamic cobweb model (see, for instance,
Hommes (1994) and Gallas and Nusse (1996)) where it is proved that prices converge to the
steady state only in the long-run. According to such consideration we use the following
equation to describe the forward-looking expectation:
µ −1
Pt e = P ƒ + γ ( Pt − P ƒ ), γ ∈ (0, 1), P ƒ = (5)
µ
ƒ
where P is the positive long-run equilibrium price and it is obviously independent of the
expectation mechanism introduced. According to such formula the agent expects a mean
between the last observed price and the price balance. The economic intuition of this choice is
that if Pt > P ƒ the agent expects the price decreasing toward its equilibrium value.

3. Choosing between expectations


We assume that the representative agent chooses between the two types of predictors
(the backward and the forward ones) as follows:
⎧ t ρ t −k
⎪∑ Pk , with probability q
Pt e = ⎨ k =0 Wt (6)
⎪ P ƒ + γ ( P − P ƒ ), with probability (1 − q)
⎩ t
e
that is Pt is a random variable and consequently the dynamical system (3) describing the
price evolution over time is a stochastic dynamical system, in fact the sequence of Pt +1 states
is a sequence of random variables that is a discrete time stochastic process obtained through
application of f µ given by (3).
We present the study of this stochastic process from the point of view of numerical
simulation. In particular we study the distribution of Pt +1 as t goes to infinity and the
distribution of Pt +1, t = 0, 1,... when q goes to zero and q goes to one. Some interesting
observations will be presented.
We plan to continue the present study giving a mathematical proof of the observations
cited previously and more in general carrying out a mathematical analysis of the stochastic
cobweb models.

e
the asymptotic behavior of series Pt and consequently Pt through f µ . See Aicardi and Invernizzi (1992) and
Ahmad et. al. (1996).

3
REFERENCES

[1] Ahmad, N., Bischi, G. I., Gardini, L., 1996. Infinite distributed memory in discrete
dynamical systems. Quaderni di economia, matematica e statistica. Facoltà di Economia e
Commercio. Urbino.
[2] Aicardi, F., Invernizzi, S., 1992. Memory effects in discrete dynamical systems.
International Journal of Bifurcation and Chaos 2, 815-830.
[3] Branch, W., McGough, B., 2005. Misspecification and consistent expectations in
stochastic non-linear economies. Journal of Economic Dynamics and Control 29, 659-676.
[4] Brock, W. A., Hommes, C. H., 1997. A rational route to randomness. Econometrica 65,
1059-1160.
[5] Chiarella, C., 1988. The cobweb model:its instability and the onset of chaos. Economic
Modelling 5, 377-384.
[6] Evans, G. V., Honkapohja, S., 1998. Stochastic gradient learning in the cobweb model.
Economic Letters 61, 333-337.
[7] Ezekiel, M., 1938. The cobweb theorem. Quarterly Journal of Economics 52, 255-280.
[8] Gallas, J. A. C., Nusse, H. E., 1996. Periodicity versus chaos in the dynamics of the
cobweb models. Journal of Economic Behaviour and Organization 29, 447-464.
[9] Holmes, J. M., Manning, R., 1988. Memory and market stability: the case of the cobweb.
Economics Letters 28, 1-7.
[10] Hommes, C. H., 1994. Dynamics of the cobweb model with adaptive expectations and
nonlinear supply and demand. Journal of Economic Behaviour and Organization 24, 315-335.
[11] Jensen, R. V., Urban, R., 1984. Chaotic price behaviour in a nonlinear cobweb model.
Economics letters 15, 235-249.
[12] Mammana, C., Michetti, E., 2003. Infinite memory expectations in a dynamic model
with hyperbolic demand. Nonlinear Dynamics, Psychology, and Life Science 7, 13-25.
[13] Mammana, C., Michetti, E., 2004. Backward and forward-looking expectations in a
chaotic cobweb model. Nonlinear Dynamics, Psychology, and Life Science 8, 511-526.
[14] Mann, W. R., 1953. Mean value methods in iterations. Proceedings of the American
Mathematical Society 4, 506-510.
[15] Michetti, E., 2000. Chaos and learning effects in cobweb models. Rivista di politica
economica 12, 167-206.
[16] Sraffa, P., 1986. Le leggi della produttività in regime di concorrenza. In P. Sraffa (ed.),
Saggi (pp. 67-84). Bologna: Il Mulino.

You might also like