A stochastic model with innovators and imitators
We examine an open model with two types of firms, called innovators and imitators, to discover the market shares of the two types. We incorporate asymmetric interactions between the two types of firms.
Let n = (n1, n2) be the state vector. We assume that the transition rates are such that
where type 1 firms grow at the innovation rate f1 plus a proportional growth rate of c1 n1. On the other hand type 2 firms, which are less technically advanced that those of type 1, grow at the rate
where /2 < /1.
The two types go out of business at the rates
j = 1, 2. We assume that type 2 firms fail more often than type 1 firms: d2 ≥ d1. By imitating type 1 firms, a type 2 firm may become type 1 at the rate
while a type 1 firm may slip back to type 2 at the rate
With these transition rates, it is easily verified that the steady-state probability distribution exists by imposing the detailed-balance conditions, provided λ12 = λ21. Let g1 = c1∕d1, and g2 = /2∕d2. We define μ := λd1d2. Then we can write the transition rates more succinctly by noting that λ21d2c1 = μg1 and λ12 /2 d1 = μg2:
while a type 1 firm may slip back to type 2 at the rate
The number of type 1 firms is governed by a negative binomial distribution, while the number of type 2 firms has a Poisson distribution,
with
and
We next examine the nonstationary solution of the master equation
and
Then, expanding
we derive coupled ordinary differential equations for the first and second moments. Unfortunately, the ordinary equations are not solvable in closed form.
We mention some special cases by imposing some conditions on the parameters in the transiton rates.
7.4.1 Case of a finite total number offirms
First, assume that g2 = 0, that d2 = d1 — c1, and that h ι is much smaller than these parameters. The first assumption means that no type 1 firms become type 2 firms. Once technically advanced, firms remain technically advanced. The second assumption means that the net dropout rate of type 1 firm is the same as that of type 2 firms. The third assumption means that c1 is much larger than f1, that is, the rate of growth of type 1 firms comes primarily from existing firms generating new firms of the same type and not from new entries. Then,
hence the sum of the numbers of firms of both types asymptotically approaches
Another consequence of these parameter values is that κ12 is constant. Suppose that it is zero at time 0. Then, it remains zero for all times. With g2 = 0 and h1 = 0, we can drop the assumption that d2 = d1 — c1 and solve for κ1 and κ2 separately, because
and
In examining the joint dynamics for κ1 and κ2, we see that the covariance κ12 acts on the time derivatives of κ1 and κ2 with opposite signs and the same magnitude.
7.5