Knowledge that Transforms

Abstract : This paper considers optimization problems in which some or all variables must take on integral values. An ability to solve such problems would be valuable in itself and would also allow handling certain kinds of heretofore intractable 'economies of scale'. An automatic algorithm for solving such problems is not given. A general approach susceptible of individual variations, depending upon the problem and the judgment of the user is presented. Two moderate-size examples are presented to illustrate the method. (Author)

The hypothesis considered is that the utility function can be written as U U[VA (qai , * **, qX**, VM(q?1 , , q.,)], where the V's are utility functions depending on quantities of different commodities assigned to different branches (e.g., food, clothing). This hypothesis implies certain empirically meaningful and interesting conditions on the price coefficients of the demand functions. These conditions are, however, not subjected here to statistical test. A price index formula is next developed for measuring branch prices (e.g., the price of food, the price of clothing) and with the use of these indices a rationale is found for the individual doing his budgeting by first deciding how to allocate expenditure among the several branches and then making independent decisions as to how best to spend each branch allocation on the commodities within the branch. That this budgeting practice is both familiar and consistent with the proposed hypothesis is taken as evidence in support of the hypothesis and, therefore, of the implications for the price coefficients in demand functions.

FEW DATES in the history of econometrics are more significant than 1857. In that year Ernst Engel (1821-1896) published a study on the conditions of production and consumption in the Kingdom of Saxony [6], in which he formulated an empirical law concerning the relation between income and expenditure on food. Engel's law, as it has since become known, states that the proportion of income spent on food declines as income rises. Its original statement was mainly based on an examination of about two hundred budgets of Belgian laborers collected by Ducp6tiaux. Since that date the law has been found to hold in many other budget surveys; similar laws have also been formulated for other items of expenditture. With the formulation of Engel's law an important branch of econometrics took its start, though it was not until our days that consumption research was placed on a sound theoretical and statistical basis. It is proper that in this centennial year econometricians should pay tribute tzo one of their most illustrious precursors. His successful attempt to derive meaningful regularities from seemingly arbitrary observations will always be an inspiring example to the profession, the more so because in his day economic theory and statistical techniques were of little assistance in such an attempt. There can, I think, be no more fitting tribute to this enlightened empiricist than a further inquiry into the subject to which he devoted much of his life's work. There is no need to go into details of Engel's analysis and of the developments that preceded it, for these matters have recently been discussed in the scholarly article by Stigler [13]. It should be enough to note that Engel was mainly influenced by two of his older contemporaries. One was the French engineer Fred6ric Le Play, who had collected budgets from households all over Europe, mostly, it seems, from humanitarian interest. Engel had been Le Play's student at the Ecole des Mines in Paris. The other main influence was the Belgian statistician Qu6telet, who was a firm propoinent of the idea that human characteristics, at least in the average, were governed by laws as definiite as those which govern