This paper reviews the growth experience of U.S. agriculture over the past two centuries in consonance with the view that growth is determined by the economic environment, which consists of the available technology, incentives, constraints, and institutions. Within this framework, the implemented technology is determined jointly with the resource allocation. The review covers the role played by resource endowment, resource flow, technical change and its factor bias, and product demand. It highlights the importance of the income elasticity of demand and the labor augmentation of the technical change. The total factor productivity (TFP) was almost nil at the beginning of the nineteenth century, increasing gradually to the point where it exhausted output growth in the latter part of the twentieth century. This pattern is consistent with the postulate emerging from this framework, where the TFP is endogenous and determined jointly with growth rather than determining it. The more recent performance of U.S. agriculture is placed within a global perspective in order to generalize the discussion.

MUCH OF THE DISCUSSION ON THE ESTIMATION of production functions is related to the fact that inputs may be endogenous and therefore direct estimators of the production functions may be inconsistent. One way to overcome this problem has been to apply the concept of duality. The purpose of this note is to point out that estimates based on duality, unlike direct estimators of the production function, do not utilize all the available information and therefore are statistically inefficient and the loss in efficiency may be sizeable. Examination of the sources of input variability suggests generalizations of currently used estimators. A SIMPLE MODEL Duality theory is a microtheory and as such its empirical implications are related to firm, rather than market, data. Our discussion is conducted within this framework and we ignore further complications that arise from the use of aggregate market data. The conceptual problems in the choice of estimators can be presented in terms of a well known simple model. Whatever other virtues more complex models possess, the problems discussed here are not resolved by the added complexity. Let the production function be

[In empirical analysis of data consisting of repeated observations on economic units (time series on a cross section) it is often assumed that the coefficients of the quantitative variables (slopes) are the same, whereas the coefficients of the qualitative variables (intercepts or effects) vary over units or periods.This is the constant-slope variable-intercept framework. In such an analysis an explicit account should be taken of the statistical dependence that exists between the quantitative variables and the effects. It is shown that when this is done, the random effect approach and the fixed effect approach yield the same estimate for the slopes, the "within" estimate. Any matrix combination of the "within" and "between" estimates is generally biased. When the "within" estimate is subject to a relatively large error a minimum mean square error can be applied, as is generally done in regression analysis. Such an estimator is developed here from a somewhat different point of departure.]

in the two sectors. Theorem 2 summarizes the discussion on this subject. The proofs of the proposition leading to Theorems 1 and 2 are illustrated graphically. In order to analyze the effect on the sign of the supply function in this economy, the ES of products with respect to their prices is expressed in terms of the parameters of the individual production functions. The expression shows clearly the augmentation effect observed by Johnson [1] (i.e., in the absence of distortion, the ES between products is larger than those which exist between factors). Finally, Theorem 3 states conditions under which the supply function is increasing, declining, perfectly elastic, or a combination of these.

AS is well known, economic growth leads, not without a feedback, to changes in the industrial composition of the economy. The seminal work of Simon Kuznets (1966, 1971) has well quantified various aspects of this process. One such aspect, which is of immediate interest to this paper, is the continuous decline in the relative importance of agriculture, a phenomenon that has already been observed by Engel (1857).' A change in industrial composition takes place through a change in resource allocation, including labor, leading to what can be referred to as occupational migration. Part of this phenomenon involves actual change of occupation by workers and part is performed by new entrants into the labor force whose choice of occupation differs from that of veteran workers. The relative importance of these two components may depend to a large extent on the rate at which the economy changes its composition as compared with the natural rate of population (labor force) growth. Independently of the form, at any point in time, the existing opportunities presumably dictate the allocation, and the more attractive the new opportunities are, the more people will be attracted. It is this premise, almost axiomatic to economists, that this paper proposes to measure. It is clear that the question of sectoral migration is of prime importance and that it bears important policy implications as well as having an analytic role. However, in spite of the importance of the subject, there does not seem to be any empirical sectoral migration equation at the macro level.2 The reason is not clear. It is possible that aggregation blurs the data and makes it difficult to obtain acceptable estimates. Whether or not this is the reason, it is clear that if such an equation is important, the data should reveal it and there must be a way to estimate it. In such an undertaking, it helps to have data with a large spread in the important variables. Such data are provided by a cross-section of countries. The plan of the paper is as follows. Section II deals with the formulation of the problem. It relies on existing and known concepts and it therefore concentrates on bringing the concepts together for the purpose at hand. Section III presents the results. A few concluding remarks appear in section IV.

regardless of the initial conditions. [8 p. 261] This means that the difference between the equilibrium value and the actual value of x is bounded for a large enough t, and that this difference tends to zero as t becomes large. So, it has been attractive to assume a monotonic convergence and to approximate the process of adjustment (or convergence) by an adjustment equation such as that used implicitly by Koyck [3] and explicitly by Nerlove [6, 7]

In estimating parameters of the Cobb-Douglas production function, assuming competition and profit maximization, the estimator to be employed depends on the specification of the behavior of the disturbance term in the production function. If this disturbance term is not transmitted to inputs, that is, if inputs are independent of this disturbance, then the least squares estimator is consistent; if the disturbance is fully transmitted to inputs, then a consistent estimator is obtained if some restrictions are imposed on the second moments of the disturbances in the system. A more general case may be specified, however, encompassing the above specifications as subcases. In this general case, the disturbance term may be only partially transmitted. If this occurs, then neither of the estimators noted above are consistent. In fairly general situations, these estimators furnish upper and lower bounds for the production function elasticity (in a one-input case) or for the sum of the elasticities (in the Q input case; Q any number). The consequences of each of these specifications, in terms of probability limits, are examined in some detail. This is carried out, first, for the one input case, and then the Q input case is discussed.