The first number of our earlier series, A Programme for Growth, carried
a notice of forthcoming papers. Five were announced but eventually only
four were published. The fifth, which was intended to deal with
consumption functions, never appeared; now it takes its place as number
one in the new series. It is not that ten years ago we had nothing to
say on the subject of consumers' behaviour. The crude estimation method
that I had used in my original (1954) paper on the linear expenditure
system gave interesting and in many respects satisfactory results, some
of which were published outside our series, for instance in Stone, Brown
and ). With this method the parameter estimates changed Rowe ( 1964 very
little after the first few iterations. Nevertheless they did change, and
with the computing resources then at our disposal we failed to reach
convergence. It was mainly for this reason that we decided to wait.