Frisch's work in all its phases exemplifies the interplay of economic theory, empirical analysis, and statistical method which peculiarly characterizes econometrics. Starting with his earliest papers, theoretical considerations have essentially the role of interpreting data. Repeatedly and in many different contexts, the need for a model, as we would say today, to enable us to learn from observations is stressed. This is particularly true when the exigencies of imperfections in data and theory lead us to statistical analysis and therewith implicitly to the admission of stochastic elements. From his empirical work, Frisch was led to examine the subtleties in the interpretation of statistical relations and the methods of statistical analysis appropriate to theoretical understanding. Such a program has inevitably meant the extensive use of mathematics, which involves two interrelated dangers: cutting the lines of communication with economists who lack mathematical training, and a tendency to value mathematical technique over economically meaningful results. In Frisch's work, the sterile Byzantinism that might be implied by these dangers is completely avoided. At all points, there is an open-minded receptivity to economic ideas derived from all sources, whether or not expressed mathematically, and the focus of all research is the underlying economic issue, not the mathematics used. This does not, however, mean any reluctance to use difficult mathematics when it is necessary to the solution. At all times, the economic problem is the master; the necessary mathematics is neither complicated for reasons of elegance and generality nor skimped for reasons of popularity. The enormous volume and variety of Frisch's work is indicated by the attached bibliography (which has been prepared by Professor Trygve Haavelmo and the University Institute of Economics, University of Oslo). This excludes (with few exceptions) his mimeographed memoranda and lectures, some of which are of the greatest importance. In reviewing his work I have been forced to be selective in a way which undoubtedly reflects my own interests and accidents of reading. No attempt has been made to discuss

IN A TYPICAL Leontief model, there is a unique method of production for each product. Hence, if a given bill of final goods is specified, the output of each industry is uniquely determined, there being no possibilities of substitution. Professor Chenery, in a contribution to a study of the industrial structure of the Italian economy ([1], Chapter II, Section E), has considered a model more general in that each commodity may be either produced domestically by a unique process or imported, which involves a drain on foreign exchange. Some, at least, of the domestic industries operate under capacity limitations. There are then alternative ways of producing a given bill of goods; choice among them is to be made on the basis of minimizing the cost of imports in foreign currency. One would expect that the choice of production and import program would depend upon the relative prices of imports. The procedure actually used by Chenery is, however, independent of these prices. The purpose of the present note is to demonstrate that his procedure is correct under a wide variety of circumstances, i.e., that in spite of the presence of a substitution possibility, the optimal choice is independent of relative prices.

[The Ramsey model of optimal capital accumulation is reconsidered under the additional restriction that gross investment must be nonnegative. An effective characterization of the optimal solution in open-loop form is obtained. It is shown, however, that in general no restriction can be placed on the number of intervals in which the non negativity constraint is binding.]

This paper is a sequel to On the Stability of the Competitive Equilibrium, I, by K. J. Arrow and L. Hurwicz. It extends the results of I in several directions. In particular, it provides a proof of stability in the large (and not merely locally) when all goods are gross substitutes; this result is found to be valid for processes where the price adjustment rate is a continuous sign-preserving, but not necessarily proportionate, function of excess demand. The paper deals both with systems where one of the commodities plays the role of numeraire and with systems where all commodities are treated symmetrically.