Section 2

Systems of Equations

Solving economic models often requires solving systems of equations. Here we

discuss solving systems of linear equations, meaning that the equations in the
system take the form
a1 x1 a2 x2 ... an xn c .
Solution Possibilities
(i)

Consider the system

2 x1 x2 4
x1 2 x2 2

A solution of this system is a pair of numbers ( x1 , x2 ) that satisfy both equations at

the same time. In this example, the solution is easily found to be ( x1 , x2 ) (2,0) .
The system (i) has exactly one solution, but this is not true for all systems. Some have
an infinite number of solutions, others have none. Consider
(ii)

3x1 5 x2 6
6 x1 10 x2 12

The second equation is simply two times the first they are not independent
equations. There is effectively only one equation in two unknowns, and any ( x1 , x2 )
that satisfies the first equation will satisfy the second. The system has infinitely many
solutions any ( x1 , x2 ) satisfying
x1 2 (5/ 3) x2

or

x2 (6/ 5) (3/ 5) x1

is a solution. The set of solutions is usually written as

5s

( x1 , x2 ) 2 , s
3

or

6 3s
( x1 , x2 ) s , .
5 5

The symbol s is sometimes called a parameter.

The fact that a system may have no solution is illustrated by the next example:
(iii)

3x1 5 x2 6
3x1 5 x2 7

If the first equation is satisfied, the second cannot be satisfied. This system has no
solution. The two equations are conflicting equations.
Drawing the equations on the x-y plane (more appropriately here, the x1 - x2 plane)
shows clearly what happens in all three situations. The two lines intersect in (i),
coincide in (ii), and are parallel to each other in (iii).

Matrix Algebra Notes

Anthony Tay

2-1

A system with at least one solution is called a consistent system. A system with no
solution is said to be inconsistent. Whether a system has one solution or many
solutions depends on whether there are as many independent equations as there are
variables to be solved.
What if there are three equations in two variables? Here we have all three
possibilities. Consider
2 x1 3 x2 18

(iv)

3 x1 4 x2 7
x1 2 x2 10

This system has no solution. Solving the last two only gives the unique solution
x1 13/ 5 , x2 37 /10 .

But this solution cannot be reconciled with the first equation:

2 x1 3x2 2(13/ 5) 3(37 /10) 163/10 18 .

Drawing the three equations in the x1 - x2 plane shows clearly what is going on here.
For there to be a solution, these three equations must intersect at the same point. In
this example, they dont.
Here is an example where the three equations do intersect at a single point.
4 x1 2 x2 3

(v)

3 x1 4 x2 7
x1 2 x2 10

System (v) has exactly one solution. Solving the last two gives
x1 13/ 5 , x2 37 /10

as in the previous case. Substituting this solution into the first equation gives
4 x1 2 x2 4(13/ 5) 2(37 /10) 15/ 5 3 so the first equation is also satisfied.
Another way of looking at this system is to observe that although there are three
equations in two unknowns, there are actually only two independent equations. Any
one of the three equations can be written as a combination of the other two (e.g., 3rd
equation is 1st minus 2nd; or 2nd equation is 1st minus 3rd, etc.) Solving any two
therefore solves the third. It is the number of independent equations that matter in
determining the number of solutions: two independent equations in two unknowns
produces a single solution.

Matrix Algebra Notes

Anthony Tay

2-2

If a three-equation two-variable system contains only one independent equation, then

effectively we will have only one equation in two unknowns, leading to infinitely
many solutions. This is the case with the following system:
2 x1 4 x2 20

(vi)

3x1 6 x2 30
x1 2 x2 10

Here the 1st equation is twice that of the 3rd, and the second is three times that of the
3rd. Therefore there is only one independent equation, in two unknowns. There are
infinitely many solutions ( x1 , x2 ) (10 2s, s) .
*******
A linear equation involving three variables x1 , x2 , and x3 , represents a plane in the
3-dimensional space. This visualization helps you to understand what can happen in
linear systems with three variables.
If you have two planes, then either the planes are parallel, are coincident, or intersect.
If they intersect, the intersection produces a line in three-dimensional space.
Therefore two linear equations involving three variables will either have no solutions
(if the two planes are parallel), or an infinite number of solutions represented by the
entire plane (if the two planes coincide) or an infinite number of solutions represented
by the line of intersection of the two planes.
The following system has infinitely many solutions represented by a single line.
(vii)

2 x1 3x2 x3 0
x1 x2 x3 1

Subtracting the second from the first eliminates x3 , and gives x1 4 x2 1 . Thus for
any x2 s , we have x1 4s 1 . Substituting into the second equation gives
x3 1 x1 x2 1 4s 1 s 5s .

We have infinitely many solutions given by ( x1 , x2 , x3 ) (4s 1, s , 5s) .

The system
(viii)

2 x1 3x2 x3 0
4 x1 6 x2 2 x3 1

however, has no solution. Dividing the second equation by 2 gives

2 x1 3x2 x3 1/ 2 . Thus if ( x1 , x2 , x3 ) satisfies the first, it cannot satisfy the second.
These two planes are parallel.

Matrix Algebra Notes

Anthony Tay

2-3

The system
(ix)

2 x1 3x2 x3 1
4 x1 6 x2 2 x3 2

has infinitely many solutions. It is easy to see that the two equations are identical.
Every point on the plane is a solution. For any x1 s , x2 t , we have
x3 1 2s 3t . The solution is
( x1 , x2 , x3 ) (s , t , 1 2s 3t ) .

*******
In order to get a unique solution to a three-equation system, there must be three
independent non-conflicting equations. The three planes must intersect at a single
point. If there are any dependence among the three equations, then we will end up
with infinitely many solutions (either a line or a plane), or if there are conflicting
equations, no solutions. We will explore this further in later sections.
Solving larger systems of equations may seem a more daunting task, but the process
of solving small systems is easily systematized. This process is called elimination
using elementary row operations. Take the system:
2 x1 2 x2 4 x3 4

eq1

3x1 1x2 2 x3 2

eq2

5x1 2 x2 1x3 7

eq3

You would generally proceed by using equation 1 to eliminate x1 from eq2 and eq3:
2 x1 2 x2 4 x3 4

eq1

eq2 minus 3 / 2 eq1

0 x1 2 x2 4 x3 4

eq2

eq3 minus 5 / 2 eq1

0 x1 3x2 9 x3 3

eq3

Then use eq2 to eliminate x3 from eq3, to get the triangular system

eq3 minus 3 / 2 eq1

Matrix Algebra Notes

Anthony Tay

2 x1 2 x2 4 x3 4

eq1

0 x1 2 x2 4 x3 4

eq2

0 x1 0 x2 3 x3 3

eq3

2-4

Finally, multiply the third equation by 1/ 3 to get x3 1 . Substituting x3 1

upwards gives x2 4 , and x1 0 .
You may have to switch the positions of your equations first. If the coefficient on the
first variable in the first equation is zero, then switch that equation with another:

0 x1 2 x2 4 x3 4
3x1 1x2 2 x3 2

3x1 1x2 2 x3 2
switch eq 1 with eq2

5x1 2 x2 1x3 7

0 x1 2 x2 4 x3 4
5x1 2 x2 1x3 7

and proceed as usual.

Notice the important role played by the boxed numbers. These numbers are called the
pivots of the coefficient matrix. If a system of equations has a unique solution it will
have non-zero pivots. Notice also that we only used three types of operations in
solving our system of equations: switching equations, multiplying by a (non-zero)
constant, and using one equation to eliminate another. These operations are called
row-operations and underlie much of matrix algebra.
********
Systems of Nonlinear Equations
We often also have to solve systems of non-linear equations, such as
(x)

xy 4

x2 y 2 8
The method for solving systems of non-linear equations is effectively also to
eliminate variables, although how best to do this depends of the system; it is a caseby-case situation.
For the system above, we might approach it in the following way: the first equation
gives x 4 / y . Thus

16 / y 2 y 2 8 16 y 4 8 y 2
which gives

( y 2 4)2 0 .

The solutions are, therefore, ( x, y) (2,2) and ( x, y) (2, 2) .

Matrix Algebra Notes

Anthony Tay

2-5

It is easy to see that solving larger systems of non-linear equations can get really
tricky. Even small systems can be tricky, and youll have to be careful of extraneous
solutions and missing solutions. For example, take
y x

(xi)

y 2 x

Perhaps the obvious thing to do here is

x 2 x

x (2 x)2 4 4 x x2

0 ( x 4)( x 1) .

So you might conclude that x 1 or x 4 , with y 1 and y 2 respectively.

However, ( x, y) (4,2) is not a solution, since this does not lie on the second
equation. [Qn: why is ( x, y) (4, 2) not a solution?]

Exercises
1.

Plot systems (i) to (vi), and (x) and (xi), each in their own diagrams.

2.

Solve the system

y x2
x2 y 2 a ,

for x 0 , y 0 , a 0 . Plot the system on the x - y plane for different values of a .

Matrix Algebra Notes

Anthony Tay

2-6