A base model matrix is defined for the largest code length of each code rate. The set of shifts {p(i,j)} in the base model matrix are used to determine the shift sizes for all other code lengths of the same code rate. Shift sizes {p(f,i,j)} for a code size corresponding to expansion factor z.sub.f are derived from {p(i,j)} by scaling p(i,j) proportionally, and a model matrix defined by {p(f,i,j)} is used to determine the parity-check bits for the f-th code.