Okay, although you don't have to use matrices, it is by far the easiest method once you understand it.
The theory behind this comes from setting out the two equations in matrix form to solve. In this case, you write both equations, one on top of the other and make sure the test variables (x and y) align with the respective counterpart in the other equation. Then you go from left to write and fill in the blanks in your matrix.
Once the matrix has been constructed, the next step would of course be to rearrange the equation so that the X matrix is by itself. This involves finding the inverse matrix of A.
To find the inverse of a 2x2 matrix, the top left and bottom right switch places, the top right and bottom left switch signs, and the whole thing is multiplied by 1 on the determinate of A.
The determinant of A [det(A)] is calculated by multiplying the top right with the bottom left and subtracting it from the bottom right multiplied by the top left.
It goes to stand that if the det(A) = 0, then the inverse of A doesn't exist.
So, all that really need be done in the first place is find det(A), set it equal to zero and solve.
See attached scan.