ABSTRACT
The recognition lag is common to all the instruments of control, though it will vary for the different targets since the time it will take to assemble the relevant statistical information will vary between domestically produced money incomes, the balance of payments and expenditures on capital investment. Nor can the recognition lag for anyone of the targets
be regarded as a single, simple, fixed period. Rough estimates may be available fairly quickly but then be liable to subsequent revisions as more and more information becomes available. The more reliable the statistical data, the more confident and bolder can be the action taken by the controllers. Thus the problem is to choose the best combination of delay before 'recognising' the need for action and of appropriate degree of boldness for the control action then to be taken. This raises an important statistical and control problem for the technical design of policy rules but one with which we have been unable to deal systematically. To do so would have required a major investigation into measurement errors.
