A Simple Linear Algebraic Approach to Capture the Dynamics of the Circular Flow of Income

The act of selling and buying as a driver of money circulation:

A fundamental concept that underlies the notion of scarcity, which is the central core of economics (e.g.,[1], page 4; [3], page 4), is the fact that there are buyers who desire to acquire the scarce goods and services in the market [4]. If no one is interested in these goods and services, then they would be irrelevant to the study of economics, even if they were scarce. The approach we propose in this paper to describe the circular flow of income focuses on the very exchange of goods and services between sellers and buyers, and attempts to capture the dynamics of this exchange from the ground up.

To describe the basic idea of this approach, let us imagine a community of three economic agents, who sell goods and services to each other in exchange for money, all in a closed economy. We refer to these three agents as A, B and C, also known as Alice, Bob and Carol (to reuse the fictional characters introduced by cryptographers Rivest et al. [2]). As a result of the business exchange taking place among them, Alice, Bob and Carol earn income. Let us denote the incomes earned by Alice, Bob and Carol up until time $t$ and expressed in monetary units as $x_{1}(t)$, $x_{2}(t)$ and $x_{3}(t)$ respectively. Between time $t$ and time $t+1$, Alice will sell goods and services to Bob who will then pay Alice a certain price $p$ for these goods and services. This price $p$ paid by Bob to Alice will come out of Bob’s income $x_{2}(t)$ at time $t$; it will be a fraction of $x_{2}(t)$. Let us denote this fraction of $x_{2}(t)$ paid by Bob to Alice during the time interval $[t,t+1]$ as $f_{12}(t)$. In a similar way, Alice will also sell goods and services to Carol between $t$ and $t+1$ and will, as a result, earn an additional income of $f_{13}(t)\times x_{3}(t)$, where $f_{13}(t)$ is the fraction of $x_{3}(t)$ paid by Carol to Alice between $t$ and $t+1$ in exchange of the purchased goods and services. Thus, the total income earned by Alice over the time interval $[t,t+1]$ as a result of her doing business with other economic agents is:

$$x_{1}(t+1)=f_{12}(t)\,x_{2}(t)+f_{13}(t)\,x_{3}(t)$$

(1)

Let us now focus on Bob as a seller of some different type of goods and services to other economic agents, namely Alice and Carol who would then have to pay some fractions $f_{21}(t)$ and $f_{23}(t)$ of their respective incomes $x_{1}(t)$ and $x_{3}(t)$ at time $t$ to Bob. Thus, over the time interval $[t,t+1]$, Bob earns a total income of:

In a similar way, Carol, as a seller, would earn a total income of:

$$x_{3}(t+1)=f_{31}(t)\,x_{1}(t)+f_{32}(t)\,x_{2}(t)$$

(3)

over the time interval $[t,t+1]$, as a result of her selling other goods and services to Alice and Bob, $f_{31}(t)$ and $f_{32}(t)$ being the fractions of the buyers’ respective incomes $x_{1}(t)$ and $x_{2}(t)$ that were paid to Carol.
 
Using matrix algebra, the above three equations can be written as follows:

$$\mathbf{x}(t+1)=\mathbf{F}_{t}\,\mathbf{x}(t)$$

(4)

where $\mathbf{x}(\tau)=[x_{1}(\tau),x_{2}(\tau),x_{3}(\tau)]^{T}$ is the income vector at time $\tau$, and the matrix:

$$\mathbf{F}_{\tau}=[f_{ij}(\tau)]_{{}_{i,j\in\{1,2,3\}}}$$

(5)

is the $3\times 3$ matrix whose entries are the above-mentioned fractions $f_{ij}(\tau)$ at time $\tau$. The superscript $\ {}^{T}$ denotes, in the entire article, the transpose of a vector or a matrix. We should note that some of the entries in $\mathbf{F}_{\tau}$ might be equal to zero if there is no exchange between the corresponding agents. In particular, according to the above linear algebraic description of income circulation, the diagonal entries $f_{ii}(\tau)$ of the matrix $\mathbf{F}_{\tau}$ will be all zeroes, because agents don’t sell things to themselves. But they can save money between one time instant and the next one. As business actors, they earn money by selling goods and services to others and, as living entities, they incur expenses to maintain themselves. The difference between the income earned and the expenses incurred would be the agents’ savings. To be more specific, let us consider one individual agent $j$ which can be Alice, Bob or Carol. The total expenses that are incurred by $j$ between $t$ and $t+1$ can be calculated in the following way using the entries in the column $j$ of $\mathbf{F}_{\tau}$:

$$(\text{total expenses})_{j}=\sum_{\begin{subarray}{c}i\in\{1,2,3\}\\ i\neq j\end{subarray}}f_{ij}(t)\,x_{j}(t)$$

(6)

Then, the difference:

Let us now redefine the diagonal entries $f_{jj}(\tau)$ for each agent $j\in\{1,2,3\}$ in the matrix $\mathbf{F}_{\tau}$ as:

An agent-based income circulation model:

The gedankenexperiment we used in the above discussion involved three agents  —  Alice, Bob and Carol. But we can easily repeat this experiment and generalize it to $n$ agents with $n>3$, in which case the final version of income circulation model becomes:

Let us now specify the context within which the above matrix model (9) would be valid:

Under the conditions specified in the above itemized list, the dynamics of the wealth vector $\mathbf{x}$ are governed by the matrix equation (9). Using the rules of matrix algebra, we can show that the wealth vector $\mathbf{x}$ can be expressed as:

$$\mathbf{x}(t+1)=\left(\prod_{\tau=0}^{t}\mathbf{F}_{\tau}\right)\,\mathbf{x}(0)$$

or, equivalently:

$$\boxed{\mathbf{x}(t)=\left(\prod_{\tau=0}^{t-1}\mathbf{F}_{\tau}\right)\,\mathbf{x}(0)}$$

(10)

where $\mathbf{x}(0)$ is the wealth vector of the economy $\mathcal{E}$ at an arbitrarily selected initial time instant $t=0$. The matrix equation (10) constitutes what we believe to be a fundamental result about the dynamics of income circulation which, to our knowledge, has never been published in the economics literature, and which can be stated as follows:

The sum $M=\sum_{i=1}^{n}x_{i}(0)$ of the coordinates of $\mathbf{x}(0)$ can be referred to as the monetary base. Since the financial sector is inexistent in $\mathcal{E}$, no money is created and, thus, $M$ remains constant over time. Using matrix algebraic rules, one can show that the sum of the coordinates of the wealth vector $\mathbf{x}(t)$ is invariant under transformations by in-homogeneous products of column-stochastic matrices. This ensures that the income circulation equation (10) is consistent with the initial setup of $\mathcal{E}$. Also, in the cases where the length of the time steps are so short that no business exchanges take place within a certain time interval $[\tau,\tau+1]$, the income circulation matrix $\mathbf{F}_{\tau}$ would just collapse to the identity matrix, as agents get to keep all their wealth to themselves during this time interval. Because the identity matrix is the neutral element for matrix multiplication, the matrix model (10) is consistent with times steps being of any size. We also conjecture that one can derive from the matrix model (10) many of the empirical observations that economists have highlighted about the issues of economic inequality, including the fact that wealth distributions in societies follow, under some conditions on the structures of the income circulation matrices $\mathbf{F}_{\tau}$ ($\tau>0$), the Pareto law.

 

About the authors:

AG and JH are two Canadian systems scientists with background in mathematics, physical chemistry and engineering science. They live in Toronto, Ontario, and can be reached at the following e-mail address: