Knowledge that Transforms

PpT HE loss of aggregate income due to the 1970 recession in the United States is widely recognized and much decried. How the loss has been distributed in society is not so well known and not extensively researched. This paper is concerned with measuring and describing the incidence of the recession on families, by income level. Historical trends in the size distribution of income have been' analyzed by Budd (1970) and Lampman (1971), and its cyclical variability has been studied by Schultz (1969), Metcalf (1972), Thurow (1970), and Mirer (1972). Most of their results suggest that macro-economic downturns increase income inequality or otherwise bear heavily on the poor and near-poor. This analysis examines micro data from a panel survey to measure the pattern of incidence of the loss of aggregate income in 1970, and finds it to be different from the effects found for past recessions. Toward the end of the 1960's, the economy was experiencing high employment along with increasing inflation. Restrictive monetary and fiscal policies along with changes in the structure of government expenditure brought about a worsening of economic conditions. In February 1969 the civilian unemployment rate stood at 3.3 per cent; it rose above 3.5 per cent in September and above 4.0 per cent in February 1970. By December 1970 the unemployment rate was 6.1 per cent. In 1970 real output declined 0.4 per cent from the 1969 level. In describing the distributional effects of these changes in macro-economic conditions, it is essential to compare what actually occurred to what would have occurred under some specified set of alternative conditions. The analytical framework of this study is a comparative statics model in which families' incomes in 1970 are compared to what their incomes would have been then if the aggregate conditions of 19671969 had continued. This approach is particularly relevant for policy purposes because it allows one to judge the distributional costs of the restrictive anti-inflationary policies of recent years.

T is an accepted proposition that there is a positive association between efficiency and experience in production. Yet, until quite recently, only industrial engineers tried to estimate and evaluate the contribution of experience to productivity, while economists have on the whole neglected this aspect of production. The growing interest in growth and the search for factors which can help to explain quantitatively the growth of production over and above the growth of the conventionally identified inputs have led economists to reconsider the problem. Arrow's pioneering article on the subject suggested that learning, which may be identified with cumulated experience, contributes to the growth of productivity.' In contrast to the engineer's approach, which usually concentrates on the effect of experience on efficiency in specific processes, the economist's approach is more general. Learning is not exclusively attributed to the individuals who make up a firm's labor force or to specific processes, though both are considered to be integral elements of the process. Rather, it is ascribed to the production unit as a multidimensional entity. It is therefore the modus operandi of the firm in each of its many facets entrepreneurial ability, technical expertise, the know-how of its labor force, layout, buying and selling, human relations, lines of command, all of them subject to continuous change which is the best testing ground of the empirical significance of the learning hypothesis. Since the acquisition of experience requires time, time or a proxy for it, such as cumulated investment and cumulated output, must obviously appear as an explicit variable in an empirical test of the hypothesis. Productionfunction analysis based on time series is therefore an obvious way of studying the quantitative effect of learning. Arrow's reference to the Swedish Horndal firm is a case in point.2 However, it is only rarely possible to find a similar example. This means that although empirical testing of the learning hypothesis can be carried out on the basis of time-series data,3 it strains the theoretical framework of the analysis. The use of cross-section data in estimating the contribution of experience to productivity is therefore an obvious alternative. Since the input and output figures necessarily refer to different firms, this way of tackling the problem is also constrained by the necessity of assuming that, except for random differences, the technologies applied by the firms concerned are identical. The requirement of intra-firm technological identity, or to put it more mildly, the similarityof-technology condition, can, in practice, be approximated under certain conditions. For example, a good approximation to the identical technology constraint can be obtained within a narrowly defined branch and for a group of firms located in a relatively small area, which maintain strong personal contact between managements so as to facilitate the inter-firm flow of information and whose labor forces are similar in background, attitude, and know-how. If the firms in such a group differ in their acquired experiences, and if these differences can be easily measured in terms of simple units, such a group may be a satisfactory testing ground for the learning hypothesis. These conditions obtain in the case of kibbutzim in Israel. They form a tightly organized group of collectives, relatively homogeneous in manpower, which has established effiReceived for publication December 6, 1971. Revision accepted for publication July 10, 1972. * This article makes use of data prepared for a comprehensive study on the Israeli kibbutz economy currently in preparation at the Falk Institute for Economic Research in Israel, and financed by the Falk Institute and the Twentieth Century Fund. The authors are indebted to Yoram Levin for his help in the preparation of the quantitative skeleton of this article. 1See Kenneth J. Arrow (1962), pp. 155-174. 2Production methods in this firm do not appear to have changed for about 15 years; productivity, however, increased continuously. 'Shifts in the production function cannot be ruled out if the period studied is long. For shorter periods, for which this objection may have less force, the number of observations may be too small to allow for statistically significant results.

IN recent years several research studies have used a data base consisting of a time series of cross-section samples. A primary reason is that panel data of this type are potentially richer in information than a single cross-section sample. To date, however, the question of how to best analyze data bases of this type has not been fully explored in any of the research. To illustrate, a number of prior research projects which have utilized this type of data base are briefly reviewed. Hoch (1962) used moving cross-section samples as the data base for estimating the parameters of a CobbDouglas production function by analysis of covariance. Specifically the data were collected on 63 Minnesota farms for the years 1946 to 1951. Hoch reported an observed difference between the least squares parameter estimates and covariance estimates and the elasticities developed from these estimates. In terms of method, the major conclusion was that the covariance model might produce less biased elasticities and marginal return estimates. Massy and Frank (1965) investigated the relationship between price changes and dealing activities on a -firm's market share for frequently purchased household and food products. Panel data covering a 101-week time period of family purchase history provided the data base for the study. However, the data were aggregated so no methodological insight could be inferred concerning the question of analyzing time series of cross-section data. Laughhunn and Lyon (1971) applied Bayesian regression in analyzing a time series of cross-section cigarette consumption data using the Tiao and Zellner (1964) approximation method. The primary methodological issue in this research was to observe differences that might exist between classical pooling and the Bayesian regression technique. Comparison of the two techniques revealed very little difference between either parameter estimates or standard errors. Schipper (1964) used covariance regression to analyze a series of cross-section samples (19541957) collected by the Survey Research Center, University of Michigan. The central focus of this study was to analyze consumer discretionary behavior particularly with respect to durable expenditures, short term debt, and discretionary saving. The major methodological finding was that several differences between the covariance regression model and individual cross-section regressions existed. Schipper suggested the individual cross-section analyses might be biased but could not prove this point since he did not use experimental data. Palda and Blair (1970) conducted an analysis of toothpaste demand by using multiple cross sections of data collected by MRCA during the period 1958-1962. One focus of their research was to investigate the potential cross-section specification bias, based on the rationale presented by Simon and Aigner (1970), that can exist because of omitted variables. An interpretation of the results led them to think that the covariance model may reduce the specification bias. This interpretation cannot be considered conclusive since the analysis was not conducted in an experimental framework. Since there is an interest on the part of economic and business researchers to use multiple cross-section sample data, this would appear to be a sufficient reason for evaluating the different methods available for combining and analyzing the samples. Earlier work in this area includes studies by Nerlove (1967, 1968). He assumed models of the form Yit = aYit-l + Uit and Yit = aYit-l + 1-Xit + Uit respectively with Uit = yi + Vit with yi and Vit uncorrelated where =o-2 = 2 +or2. The estimation methods used were OLS, generalized least squares utilizing known p (p = o-A2/o-X2) analysis-of-covariance estimates with cross-sectional effects only, two-round estimates based on an estimated value of p, and maximum likelihood estimates. Generally, Nerlove's findings indicated that generalized least squares (if p is known) produces good esti-

D ESPITE failure of rich countries to grant any new tariff preferences to less developed countries (LDC's) 1960's, export earnings of manufactures by LDC's grew by more than 10 per cent per year during decade.' This phenomenon seems related to !the expansion of multinational firm. Scattered bits of evidence suggest that a large fraction of exports of manufactures from LDC's are accounted for by such firms. Between 1965 and 1968 annual exports from LDC's by foreign affiliates of United States manufacturing firms rose from 700 million dollars to 1.4 billion dollars.2 Between 1957 and 1966 Latin America's annual exports of manufactures rose from 709 million dollars to 1,613 million dollars, and subsidiaries of United States firms accounted for 65 per cent of this increase of 804 million dollars.3 IBM is said to have been largest single exporter of manufactures from both Argentina and Brazil 1969.4 As is well known, there is a large amount of literature on question of impact on currently less developed countries of large increase their exports of primary products during 19th century. We do not have much evidence on impact of contemporary investment by foreign manufacturing firms LDC's.5 The Pearson Commission (1969, p. 104) said in absence of detailed empirical studies, it is difficult to pass a definitive verdict on precise size of contribution which foreign investment has made to development. 6 Vernon (1971, p. 181), discussing extent to which foreign firms introduce into developing countries production techniques that are excessively capital intensive, says the actual facts are, as usual, obscure. There are no comprehensive data on degree to which multinational enterprises adapt their production processes to conditions of less-developed countries, and scarcely any data on comparative adaptive actions of local competitors. The next two sections present data on a sample of Japanese and United States firms South Korea, and final section discusses benefits and costs to South Korea of these foreign investments.

T HERE have been a number of attempts to estimate the relationship between schooling and earnings of individuals, but the most common feature of these studies has been severe data limitation which, in turn, has dictated how the analysis could proceed. Perhaps the most serious restriction imposed by the data has been the assumption that earnings relationships are the same across the nation or, at least, across very sizable aggregations of states. This paper examines the viability of such assumptions by looking at differences in earnings functions among smaller, more homogeneous labor markets. This res-earch reveals large differences across labor markets in the returns to human capital and indicates that much of the observed difference in regional and racial earnings results from structural differences in earnings functions.