Knowledge that Transforms

We had divided the demand for money during a period of time according to the method used to satisfy that demand: sale of goods, sale of securities, and keeping money already held. Obviously the first two constituents involve flow-analysis in the usual sense money, goods, and securities must change hands during the period to satisfy the demand for money. Nor is it the intention of the demanders to

FOR many economic problems it is important to know how business expectations behave. Are they external to the system or are they related to current or past economic events? And, if the latter, how are they related? Expansion of our knowledge in this area meets with rather formidable difficulties. If one proceeds on the basis of a questionnaire, the answers procured may fail to capture the real expectation of a business manager. His answer may be colored by the use that he believes will be made of the information, or he may be annoyed at the intrusion of an irrelevant question, or at its difficulty, and give off-the-cuff replies. Objective data would be of real usefulness, fragmentary or incomplete as they might be. For more than a decade corporations were required to pay federal taxes which implicitly called for a profit forecast of varying lengths of time. Under the capital-stock tax and declaredvalue excess-profits tax, repealed by the Revenue Act of I945, the more precisely could income be forecast, the lower would be total tax liabilities of the firm. A direct monetary reward was granted for precision of estimate. The data these taxes provide are unique in that the forecaster had to place a bet on his guess. Subject to the qualifications discussed later, these data for the period I933 to I94I seem to bear out the conclusion that expected future income was intimately related to current income. A change in current income led to a change in expected future income of almost an equal amount (94 per cent of the change in current income).

During the past twelve years most of the theoretical discussions of the problem concerned long-run effects of income changes which are not considered in this paper. The Keynesian thesis about the short-run relationship between changes in income and saving appears to have been accepted by most students.2 Empirical evidence for the validity of the short-run relationship may be sought, and was usually sought in the past, on the aggregative level, by comparing year-to-year changes in national income, aggregate consumption, and saving. The same relationship that prevails between aggregates may also be found, however, as expressly stated by Keynes, between changes in income and consumption of individual income receivers (or groups of income receivers). It is the latter problem that will be raised here: do recent studies of the financial behavior of families shed light on the relation

THE main purpose of this article is to record and analyze the results of some investigations into United Kingdom consumption data over the period I870-I938. The discussion is arranged in three sections consisting respectively of a brief survey of the economic and statistical problems involved, a summary of the data used and the results obtained, and an appraisal of the significance and value of these results. It should be emphasized at the outset that section I is little more than a survey of the work done in this field by others.' The main contribution which it is hoped will be made by this article is to extend these ideas and formulations over a longer time-period than has previously been attempted with United Kingdom data.