Knowledge that Transforms

T HIS paper sets out to analyze the empirical relationship between the structure of industrial exports and the level of income per capita across 30 countries. To do so, industrial exports are classified according to (a) the factor intensity of their production functions-for the purpose of providing an indirect test of the factor proportions theory of international trade-and (b) the income elasticity of the demand for various goods-for the purpose of testing part of Linder's (1961) hypothesis. Tariffs, subsidies, differential exchange rates, and other bafriers to trade are of crucial importance in the determination of the actual flows of goods between countries. A single variableincome per capita-therefore cannot by itself explain the structure of trade. However, since per capita income reflects the effects of a variety of economic processes and is usually regarded as an indicator of a country's level of development, I believe it worthwhile to analyse the empirical links between the structure of industrial exports and a country's level of development as measured by its per capita income. The sample of 30 countries with a per capita income above U.S. $500 in 1970 includes all the industrial countries and most of the more advanced developing countries. 1,2 I. The Factor Proportions Theory

George A. Akerlof, Ross D. Milbourne, New Calculations of Income and Interest Elasticities in Tobin's Model of the Transactions Demand for Money, The Review of Economics and Statistics, Vol. 60, No. 4 (Nov., 1978), pp. 541-546

This paper presents a historical model of non-financial corporate cash demand. Although 1947-1971 data are used here, in another article the writer has applied the same formulation to earlier years. The main thrust of the argument is that (1) Very large firms have had a different demand-forcash function from other firms. (2) When the business population changes substantially, relative increases or decreases in numbers tend to be greater where entry and exit are easier, among smaller firms. Thus, business structure has changed with business population. (3) Aggregate corporate cash holding is therefore influenced over the long run by changes in corporate population, as well as by changes in aggregate receipts, interest rates, and cyclical factors. Empirical estimates indicate that the sector's postwar holding of cash, relative to receipts, was substantially less than it would have been if the composition of the corporate population had not changed as the number of firms exploded after 1947.

A number of empirical studies would seem to cast doubt upon the neo-classical view of the supply side of an economy. Estimates of Cobb-Douglas production functions and the apparent procyclical movement of real wages would seem to contradict the assumption of diminishing returns to labor and the implication of neo-classical theory that factors of production are paid the value of their marginal product.' Recently, Lucas (1970) and Sargent and Wallace (1974) have outlined a theory that could reconcile these statistical results with a basically neo-classical view of supply relationships; essentially, they suggest that these paradoxical empirical observations are the result of specification error, of aggregating labor and wages over straight-time and overtime work shifts. This paper reports an attempt to evaluate empirically both the allegations against the neo-classical view and the Lucas-SargentWallace response within the context of the Canadian economy. In the first section, evidence concerning the diminishing returns to aggregate Canadian labor and the cyclical behavior of average real Canadian wages is presented. In the second, the effect of disaggregating labor is discussed. Conclusions will be found in the last section.

AN important unresolved question in the analysis of capital investment is whether the deterioration of physical capital occurs at a constant exponential rate. (For detailed discussion see both Jorgenson (1971) and Feldstein and Rothschild (1974).) The empirical evidence on physical capital depreciation patterns is inadequate to refute any proposition about the actual decay of capital inputs. On the one hand, analyses of acquisition prices of new and used automobiles by Wykoff (1970) and by Cagan (1965) and of used farm tractors by Griliches (1960) uniformly concluded that the geometric depreciation function adequately characterizes the true depreciation function. On the other hand, Hall (1971) and Feldstein and Rothschild (1974) found statistical grounds for rejecting the null hypothesis of an exponential depreciation function. The major purpose of this paper is to apply the statistical test procedure on the geometric depreciation function developed by Hall (1971) and Jorgenson (1971) to data on Japanese fishing boats. The published Japanese data on the insured value of fishing boats against total loss are adjusted to reflect the assessed market valuation for insurance purposes. These adjusted values are used as proxies for the acquisition prices of new and used fishing boats. These proxy variables can be justified because it is common practice for the insurance companies to minimize the difference between the insured value of capital items and their market or actual replacement value. This is accomplished by annual adjustments of insured values for physical deterioration and technical obsolescence (Lee, 1973). These data form a unique set for comparison with the acquisition price data on used automobiles frequently employed in previous studies. To date, most attention in the literature on capital depreciation has been paid to consumer durables, particularly to automobiles, with the exception of Griliches (1960). By comparison, fishing boats are capital inputs to fishery production activity. In Cagan's (1965) expression, the demand for capital inputs is less likely to be influenced by ephemeral fads and will reflect greater emphasis on innovations of enduring importance than will the demand for consumer durables. In addition, since the assessed market valuation of new boats for insurance purposes is believed to reflect satisfactorily the actual value of new boats, the data should reveal some information on that portion of the lifetime depreciation of a capital asset that takes place during the first year of its life. By contrast, the data constraints of previous studies allowed capital depreciation analyses only for vehicles that were more than one year old. This was an unfortunate and significant omission.' Though the Hall model (1971) is fundamentally under-identified, as will be discussed in detail later, it is still possible to statistically test several typical assumptions regarding the nature of depreciation and embodied technical change that are commonly seen in economic literature. While the results of the test are specific to the Japanese fishing fleet, interpretation of these results along with the results of previous studies should give some indication whether the typical assumptions are empirically Received for publication September 14, 1975. Revision accepted for publication June 23, 1977. * Research Triangle Institute. An earlier version of this paper was presented at the Third World Congress of the Econometric Society, Toronto, Canada, in August 1975. Grateful appreciation is extended to W. Kenneth Poole, Jerome A. Olson, and Allen K. Miedema at Research Triangle Institute for discussion and comments on earlier drafts and to Joanne Turner Rogoff of Research Triangle Institute for her editorial assistance in the preparation of this article. The author also thanks an unknown editorial referee for invaluable comments and criticisms. I Hall (1971) measured the depreciation of half-ton pickup trucks relative to 1-year-old trucks because the prices of new trucks were not available from his sources. Wykoff's (1970) study of acquisition prices of new and used automobiles reveals a sharp drop between the price of new cars and the price of used cars. Jorgenson (1971) contends that Wykoff's finding is attributable mainly to the inadequacy of the list prices of new equipment to reflect the prices paid at actual transactions.

THE effects of factor mobility on the regional differences in factor returns and on regional growth have been central to the theoretical analysis of regional' growth. Those who believe that economic growth of a region is dependent on the growth of inputs (either because aggregate demand for the region's output is not a constraint as in the neoclassical growth models or because a demand type model of the post-Keynesian variety is not sufficient to explain the growth of an open region) cannot ignore the role played by factor mobility. The interest in the effects of factor mobility on regional growth through its effect on the rates of growth of inputs is more than academic. Appropriate regional growth policies cannot be designed without prior determination of the role played by input growth and the variables which influence those growth rates. Yet,' empirical studies of these issues are scarce. So far only three studies have investigated the relationships between input growth and regional growth, and factor returns and input growth. The seminal work has been that of Borts and Stein (1964), who studied the response of factor mobility to factor price differentials and the relationship between output and the aggregate capital-labor ratio. However, empirical results are inconsistent with standard neoclassical growth theorems... (Smith, 1974, p. 166). These inconsistencies are attributed by Smith (1974, 1975) to the use of data on one sector (nonagriculture) without considering the possibility of intersectoral factor movements. Smith has argued that one can either consider total regional output and ignore intersectoral factor movement (1974), or consider sectoral output and allow for intersectoral factor mobility (1975). Although the empirical results reported by Smith (1974) for his two sector model are interesting,they are of limited value for our purpose since the parameters estimated and tested are those of the reduced form and no indication of the explanatory power of those equations are reported. From those results one cannot assess the contribution of factor mobility to regional growth nor the value of the model in explaining the variations in growth rates among states. The coefficients of determination are reported in the second study (1975) for the reduced form equations; however, the same limitations on the value of the results are encountered. Furthermore, the explanatory power of the reduced form equations are distressingly low. In this article we report on the results of estimating a simple neoclassical model of regional growth. The estimated model is validated by comparing its simulations with observed values of the variables and their rates of growth between 1963 and 1973. We then use the model to derive the long run implications with respect to growth rates of output, output per worker and the level of output per worker. These are obtained from the dynamic simulation of the model over a period of time. The results of these simulations reveal a strong tendency for the rates of growth output and output per worker to converge. No such convergence is obtained for the level of output per worker.