Knowledge that Transforms

Female headship among black families long has been more pronounced in the United States in comparison with other ethnic groups. E. Franklin Frazier's classic study of the black family in the 1930's placed a distinct emphasis on the disproportionately high number of urban Negro families with women heads. Frazier's work suggested that throughout the pre-World War II period almost one-quarter of black families were headed by women. In the mid-1960's, female headship among black families was the subject of Johnson Administration policy planner Daniel Moynihan's notorious characterization of the black family as enmeshed in a tangle of pathology. But while the subsequent debate between disciples of Moynihan's pathology-disorganization perspective and the proponents of the strength-resiliency perspective raged, the proportion of black families with female heads has risen markedly. The proportion climbed from slightly less than 25 percent in 1965 to an astonishing more than 40 percent by 1980. Female headship also has grown among white families, but the rate of increase has not approached that among blacks. Between 1965 and 1980, the percent of white female-headed families rose from 9 percent to close to 12 percent.

One of the validity of a scientific hypothesis its forecasting accuracy. Thus one of the hypothesis that the market can be used to allocate manpower to defense, as to any other occupation, its ability to predict voluntary enlistments. In our earlier paper, a simple model of accessions and enlistments to the U.S. armed forces was estimated, the conclusion being that the allvolunteer force (A VF) is an experiment in market economics which, far from having failed, has not yet been put to the test (McNown et al., 1980, p. 130). Although one may not be particularly sanguine about the prospects for testing the market experiment itself, one can at least expose the hypothesis to further scrutiny. This paper reports the results of an accuracy analysis of forecasts generated by the model, and, since the results on the whole are encouraging and the topic remains of considerable interest to the public and to policymakers alike,1 also presents updated estimates of the personnel supply elasticities. I. A Military Personnel Supply Model

Milton Friedman's famous essay on methodology (1953, pp. 3-43) has been presented as an instrumentalist argument for instrumentalism (Boland, 1979). Although Friedman has stated that such characterization of his methodology is entirely he also says (1979) that his views on methods and the philosophy of can be aligned with those of the philosopher Karl Popper. Many students of the philosophy of science would be possibly shocked by such a claim, since Popper has so often criticized and rejected both instrumentalism and Logical Positivism (1965). Thus, against such background, we present our examination of Friedman's claim. Evidence which we gathered to determine the extent of the relationship between Friedman's methodology and Popper's view of science will be presented. It will be argued that to a degree Friedman's claim can be supported. The major point is that Friedman identifies with Popper for two reasons: 1) Both Friedman and Popper reject Positivism; 2) Friedman and his followers tend to argue indirectly in advancing a view by criticizing its alternatives. Presumably, we either accept Positivism (for example, Paul Samuelson, Carl Hemple, Rudolph Carnap, among others), or we do not. Obviously, this line of argument considers only two among a possibly larger number of options (a confrontation of competing views, as were, as found in Friedman's view of hypothesis testing). A minor point is that Friedman does not see Popper dismissing instrumentalism completely. In this he is correct, since Popper only criticizes instrumentalism as a philosophy of science, while accepting as a methodology for social policy. Except for a brief statement that Factual evidence can never 'prove' a hypothesis, can only disprove it (1953, p. 9), nothing in Friedman's essay seems to depend on Popper's philosophy of science. So, if there is an alignment of Friedman's view with Popper's, will have to be found separately from the famous essay. Thus, we examine the prospect with reference to his academic and research backgrounds (including examples from his work), and to his association with Popper. A corollary purpose of our essay is to indicate how Friedman's view of the methodology of positive economics depends on his acceptance of some (but perhaps not all) of Popper's philosophical discussions of the nature and purpose of science. As we indicate, Friedman's early orientation in methods had no immediate background in established economics. That the main and crucial elements of Friedman's work came from outside may be gleaned from a review of these elements and their antecedents. This emphasis reveals the following: 1) that Friedman has been incorrectly identified with Logical Positivism (Warren Gramm, pp. 169, 171, 175; Samuelson, 1963, pp. 82-83), a doctrine about the unity of science as formulated in Vienna in the 1920's; and 2) that Friedman is much less aligned than commonly presumed with the views of his Chicago mentors. In Friedman's economics, theories are partly arguments for alternative policies (and/or social reform). An interdependent system of analytic constructions (rather than isolated construction in economics) is used to generate hypotheses and for predicting the effects of alternative policies. Contrary to the conventional views of statistical induction and logical deductions from known true assumptions, as dealt with elsewhere (Boland, 1979, pp. 512-13; 1981), the purpose of economic theory for Friedman is prediction for purposes of testing and evaluating alternative policies. These policies over Friedman's career *The University of Florida and Simon Fraser University, respectively. We thank John Chant. Stephen Easton, Herbert Grubel, Zane Spindler. and James F. O'Conner for comments on an early draft. Milton Friedman was cooperative in answering questions we raised. We alone are responsible for the final product.

In a recent paper, Lawrence Summers (1981) analyzes alternative taxes in a life cycle growth model. This paper extends his analysis to include bequests and inheritances. This inclusion is worth undertaking for several reasons. First, several recent empirical studies have suggested that bequests may account for an important share of the capital stock (Laurence Kotlikoff and Summers, 1981). Second, it is important to explore whether the comparison of taxes is significantly affected when a bequest motive is introduced. Third, the treatment of bequests under a consumption tax is controversial. The endowment rationale calls for treating a bequest as taxable consumption by the deceased (Peter Mieszkowski, 1980); the wage rationale calls for exempting bequests if such exemption raises the steady-state capital intensity and real wage (my 1980 article). This paper confines itself to positive analysis of steady states. Transition paths, and normative implications, are explored elsewhere (see my 1982 paper).

It has long been recognized that technological change is one of the major forces influencing an industry's market structure. Karl Marx stressed this fact over a century ago. Like Marx, many economists, including Arthur Burns (1936) and John Kenneth Galbraith (1967), have been convinced that technological change tends to increase plant sizes and the level of industrial concentration. Others, like John Blair (1972), have argued that, although such a trend existed in the past, it has been reversed since World War II because of a fundamental change in the nature of technological advance whereby centralizing technologies have been displaced and superceded by decentralizing technologies. In recent years, there has been a revival of interest in the effects of technological change on market structure. Richard Nelson and Sidney Winter (1978) have formulated a computer model and Richard Levin (1980) has estimated an econometric model, both aimed at representing these effects. In general, these models seem to suggest that a relatively rapid rate of technological change in a particular industry is likely to result in a relatively high level of concentration. However, these authors are careful to point out that their results are preliminary and tentative. Although the effects of technological change on market structure are of fundamental importance to both economic analysis and public policy, it is surprising how little systematic study has been devoted to them. We know little or nothing about the effects of the various process and product innovations that have occurred in recent years in various industries. And we have very little information concerning the relationship between the rate of technological change in a particular industry and the changes in the industry's market structure. My purpose in this paper is to try to begin filling these notable gaps.

The economic turmoil of the 1970's resulted in record postwar increases in inflation, unemployment, and nominal interest rates. Yet, it was declines that were hardest to explain. The average real value of a share of common stock plummeted. Productivity growth evaporated. Real interest rates, measured as the spread between nominal interest rates and the inflation rate, turned negative (see Table 1). Over the past decade, an explosion has occurred in the amount of attention paid to the relationship between nominal and thus real interest rates and expected inflation rates. A good deal of this effort has been directed toward empirical analysis of the Fisher neutrality hypothesis that nominal rates respond one for one with expected inflation rates. Most empirical tests of the Fisher hypothesis have been bivariate; interest rates were regressed on a constant and on actual or expected inflation measures. Estimates of the impact of inflation on interest rates in these and even in extended models are often significantly below one, are often statistically imprecise, and tend to be unstable over time.' The middle column of Table I reflects this. The coefficient of inflation on nominal interest rates there drops from 0.78 to 0.59 in the latter 1970's. Another branch of work on nominal interest rates has concentrated on the institutional impediments to the Fisher hypothesis. Robert Mundell and James Tobin demonstrate that nominal rates change by a smaller amount than the expected inflation rate does when a real balance effect exists and money pays no interest. Michael Darby and Martin Feldstein, on the other hand, argue that nominal rates should exhibit a greater-thanunity response to expected inflation due to the nature of U.S. income tax laws.2 Here I estimate the relation between interest rates, expected inflation, and real forces. Such estimates are required for evaluating the effects of inflation on saving, investment, the distribution of income, and the redistribution of wealth. To generate estimates of the net impact of these various factors on interest rates and to avoid the identification and simultaneity problems masked, but not often remedied, by instrumental variables techniques, I employ only exogenous regressors. To buttress the argument, I examine the individual links in the chain which are summarized in the reduced form. The novel aspect of the model presented in Section I is its addition of aggregate supply shocks to the determination of interest rates. I trace the reduction in the supply of complementary factor inputs in the 1970's to a decline in the demand for capital and therefore in real interest rates. The inclusion of this supply force along with expected inflation allows one to distinguish between two, offsetting effects on interest rates: the depressing effect on real rates through lower investment demand and the elevating effect on nominal rates of higher expected inflation. The results in Section II not only strongly support the economic and statistical importance of supply shocks on real interest rates, but also resolve some longstanding interest rate puzzles. Allowing for the impact of factor supply produces a significant estimated response to expected inflation for a sample that ends prior to late 1960's. The magnitude *Assistant professor, School of Business Administration, University of California-Berkeley. I would like to thank George Akerlof, Robert J. Gordon, Robert A. Meyer, Joe Peek, Janet Yellen, the members of the Economic Analysis and Policy seminar at Berkeley, participants at the NBER Conference on Inflation and Financial Markets, and anonymous referees for helpful comments. Data Resources supplied data and computational services. Financial support was provided by the Berkeley Program in Finance and NSF grant SES8109093. Linda Pacheco supplied able research assistance. Errors of omission or commission are my responsibility. 'See William Gibson, David Pyle, John Carlson, and Thomas Cargill and Robert Meyer. 2Maurice Levi and John Makin derive the reducedform effect of expected inflation on nominal interest rates as a function of the structural parameters.