- Purchase/rental options available:
History of Political Economy 35.1 (2003) 77-104
[Access article in PDF]
Milton Friedman and the Emergence of the Permanent Income Hypothesis
For two decades, in the early stage of his professional career, Milton Friedman worked on consumption theory, exploring the possible explanations of the relationship between consumption and income. The well-known “permanent income hypothesis,” often connected with his groundbreaking book A Theory of the Consumption Function (1957), is the pinnacle of Friedman's achievement in this field. Nonetheless, the permanent income hypothesis also raises the issue of defining and measuring “permanent income,” which unavoidably correlates with performing the same tasks on “income.” The modern definition of income is owed to John R. Hicks's book Value and Capital (1939), which identified income as the budget constraint of consumption behavior. But Hicks needed more specific definitions for empirical purposes. He claimed in his book that there are still great discrepancies between the theoretical definition and the empirical one and that existing theories could not help us with those discrepancies—in other words, there remains a [End Page 77] measurement problem. Thus he concluded that these income theories are “bad tools, which break in our hands” ( 1946, 177).
While Hicks attempted to find a satisfactory empirical counterpart for a theoretically sound definition of income, other economists were conducting empirical studies of income structure. Among them, Simon Kuznets was the most notable. Apart from a series of works in the 1930s and 1940s on the measurement of national income, Kuznets started a study comparing incomes of different professions in 1933 by using data for 1929–32. He completed a draft in 1936, but Friedman took up the work in 1937 and provided a more detailed statistical analysis.1 The result was published in 1945 and was titled Income from Independent Professional Practice, by which Friedman earned his doctoral degree at Columbia University in 1946.
Income from Independent Professional Practice gives an account of income structure: income is composed of permanent, quasi-permanent, and transitory components. It also marks the first of three stages of Friedman's research on the permanent income hypothesis. Each stage corresponds to a different concept of permanent income. In the second stage, identified by A Theory of the Consumption Function, Friedman continued to work on the dichotomy of measured income but extended it to the account of consumer behavior, that is, the permanent income hypothesis. A formal statement of the permanent income hypothesis was given, several empirical models were built, and tests for various types of data were proposed. Consequently this book is regarded as “one of the masterpieces of modern econometrics” (Blaug 1998, 69). Finally, Friedman (1963b) extended the treatment of his later book (1957, chap. 5), which had adopted Philip Cagan's adaptive expectations hypothesis (Cagan 1956), to the case of time-series data. By doing so permanent income is made equivalent to adaptive expected income. It became the most popular version of permanent income.
Several surveys have explored the permanent income hypothesis extensively, namely, Mayer 1972, Hirsch and De Marchi 1990, and Hynes 1998. Thomas Mayer's book is the most well known among them. Mayer indicates three definitions of permanent income in Friedman's work:
M1 Permanent income is whatever the household thinks it is. [End Page 78]
M2 Permanent income is equal to the household's wealth times the relevant discount rate.
M3 Permanent income is an exponentially weighted average of past incomes plus a trend.
Nonetheless, the main task of Mayer's book is to scrutinize the existing tests with respect to various data in order to appraise the permanent income hypothesis. But the interplay between data, models, and theoretical definitions that appears in the permanent income hypothesis received less attention.
Both Hirsch and De Marchi 1990 and Hynes 1998 see Friedman's permanent income hypothesis as attempting to give a more empirically adequate account of consumption behavior, in the sense that the permanent income model is made “operational” by containing certain auxiliary statistical restrictions such that the permanent income hypothesis is capable of being subject to...