Knowledge that Transforms

sector) disaggregation, the change in the level effect accounts for only 0.35 (0.41) percentage point of the 2.24 percentage point deceleration in labor-productivity growth between the 1948-65 and 1973-78 periods. The aggregate rate and level effects are not very sensitive to the degree of disaggregation (12versus 60-sector). Although some large level effects are evident within and across major industrial divisions, they often cancel out in the aggregate. However, the level effect from the shift out of farming is quantitatively large and is close to the aggregate level effect using either 12-sector or 60-sector disaggregation. Although one can approximate the aggregate level effect by looking only at that for agriculture, one should note that other fairly large level effects exist both within and across major industrial divisions other than agriculture.

Automotive News, 1967 Almanac (Detroit: Slocum Publishing Co., 1967). Bucklin, Louis P., Competition and Evolution in the Distributive Trades (Englewood Cliffs, New Jersey: Prentice Hall, 1972). Federal Trade Commission, Annual Line of Business Report, 1974 (Washington: FTC, 1981a) processed. , Annual Line of Business Report, 1975 (Washington: FTC, 1981b) processed. Guth, Louis A., Robert A. Schwartz, and David K. Whitcomb, Use of Buyer Concentration Ratios in Tests of Oligopoly Models, this REVIEW 58 (Nov. 1976), 488-491. Jeifreys, James B., The Distribution of Consumer Goods (New York: Cambridge University Press, 1950). Lilly Digest (Indianapolis: Ely Lilly and Co., 1976). Martin, Stephen, Advertising, Concentration and Profitability: The Simultaneity Problem, BellJournal 10 (Autumn 1979), 639-647. National Petroleum News, Fact Book 1976 (New York: McGraw Hill, 1976). NCR, Expenses in Retail Business (Dayton, Ohio: NCR, undated but about 1966 and about 1976). Strickland, Allyn D., and Leonard W. Weiss, Advertising, Concentration, and Price-Cost Margins, Journal of Political Economy 84 (Oct. 1976), 1109-1123. U.S. Department of Commerce, Bureau of Economic Analysis, The Detailed Input-Output Structure of the U. S. Economy: 1972 Volume 1, The Use and Make of Commodities by Industries (Washington, D.C.: Government Printing Office, 1979). U.S. Treasury Department, Internal Revenue Service, Statistics of Income Source Book, 1972, processed. Ward, T. S., The Distribution of Consumer Goods. Structure and Performance (New York: Cambridge University Press, 1973). Weiss, Leonard W., Geographic Size of Markets in Manufacturing, this REVIEW 54 (Aug. 1972), 245-257. , Case Studies in American Industry (New York: John Wiley and Sons, 1980).

T HE dietary choice of households near subsistence levels of nutrient intake is one of obvious policy importance. In many countries, such as Bangladesh, national goals are set in terms of nutritional intake and there is heavy intervention in the markets for foods. However, little is known about the manner in which food preferences vary with food expenditure and nutrient intake. The design of efficient programs to aid nutritionally deficient households in attaining minimal levels of nutrient intake requires information on all ownand cross-price elasticities for both target and non-target groups. The net effect of a food price subsidy on the consumption of food nutrients cannot be predicted without knowledge of the complete elasticity matrix. Results presented below demonstrate that substitution effects can be so strong that the subsidization of certain foods quite often reduces nutrient consumption. In this study, demand equations for nine foods which allow for extremely flexible consumer price response are estimated from the individual budgets of 5,750 rural Bangladeshi households. Estimation at the household level is preferred because it more readily permits the incorporation of household composition variables into the demand analysis, such as household size, occupation and employment status, that are typically lost in aggregation. There is also a greater range and variation in expenditure levels than found in grouped data. This is of particular importance in the study of nutritional well-being as it is the poorest households which are of special interest. Moreover, the household sample provides sufficient degrees of freedom to estimate a simple varying parameter model which requires the estimation of 270 parameters. Previous econometric analysis of income-class specific dietary choice has been limited and not altogether satisfactory. Pinstrup-Anderson, de Londono and Hoover (1976) estimated complete sets of price elasticities for different income strata using Frisch's scheme in order to study the impact of changes in relative prices on nutrient consumption. Their results are suspect because of the assumption of want independence necessary for this methodology to be valid. Alderman and Timmer (1980), who were also concerned with studying the relationship between food price policy and nutrient intake by income classes, econometrically estimated separate price coefficients for each income group by including slope dummy variables in their demand equations for rice and cassava in Indonesia. The inclusion of these dummy variables revealed surprisingly large differences in compensated price response across income groups. Although their results support the notion that poorer households respond differently to prices than the rich, the limitations of their data constrained them to consider only two foods and to specify changes in price response which are discontinuous with respect to income. I The formulation and estimation of the food demand equations is discussed in section II below. Section III presents the results of the estimation and discusses the nutritional implications of movements in relative food prices and other exogenous variables. Section IV summarizes our findings.

E MPIRICAL results reported in economics journals are selected from a large set of estimated models. Journals, through their editorial policies, engage in some selection, which in turn stimulates extensive model searching and prescreening by prospective authors. Since this process is well known to professional readers, the reported results are widely regarded to overstate the precision of the estimates, and probably to distort them as well. As a consequence, statistical analyses are either greatly discounted or completely ignored. This unfortunate equilibrium in the market for information is a result of the current econometric technology, which generates inferences only if a precisely defined model were available, and which can be used to explore the sensitivity of inferences only to discrete changes in assumptions. The reporting of a complete sensitivity analysis is ruled out therefore first, because the econometric theory which takes models as given would be rendered explicitly inadequate if the sensitivity analysis were reported, and, second, because the econometric technology, if used to explore sensitivity issues, would generate vast numbers of estimated models that journals are rightfully reluctant to print. It is the purpose of this article to discuss an alternative econometric technology that could increase the value of our profession's limited data resources. The basic assumption underlying this technology is that no econometric model can be taken as given. Because there are many models which could serve as a basis for a data analysis, there are many conflicting inferences which could be drawn from a given data set. If this fact of life is acknowledged, it deflects econometric theory from the traditional task of identifying the unique inferences implied by a specific model to the task of determining the range of inferences generated by a range of models. We propose that researchers be given the task of identifying interesting families of alternative models and be expected to summarize the range of inferences which are implied by each of the families. When a range of inferences is small enough to be useful and when the corresponding family of models is broad enough to be believable, we may conclude that these data yield useful information. When the range of inferences is too wide to be useful, and when the corresponding family of models is so narrow that it cannot credibly be reduced, then we must conclude that inferences from these data are too fragile to be useful. This contrasts greatly with the reporting schemes currently used by individuals. As a profession, however, we do suspend judgment on econometric results until they hold up to inspections by other researchers using other models. The advocacy process we use to accumulate professional opinion is therefore aimed in the same direction as our proposals, but the path we recommend is much more direct and the outcome is much more clearly stated. A simple introduction to this alternative econometric technology is given in section I of this paper. In writing this section we have attempted to communicate the main ideas as concisely as possible. As a consequence, there is no reference to any sophisticated statistical theory and especially no mention of the Reverend Thomas Bayes. For a more complete statement as well as theological fanfare, consult Leamer (1978). The proper test of our proposals is whether they are useful in practice. We believe that researchers will find them to be efficient tools for discovering the information in data sets and for communicating findings to the consuming public. In an effort to make clear the value of these techniques we present two examples in section II. These methods are not without their own problems, the most serious of which is their concentration on the point estimation problem and their neglect of hypothesis testing or interval estimation. The basic approach to studying and reporting the fragility of estimates which we describe in this paper can be readily extended to studying and reporting the fragility of t-values, though computational difficulties do arise. Received for publication June 15, 1981. Revision accepted for publication August 2, 1982. * University of California, Los Angeles, and Harvard University, respectively. Research supported by NSF Grant SOC78-09477. Comments of the referees have helped to improve both the content and the exposition. Thomas Wolff is thanked for able research assistance.

Many commentators believe that the phenomenon of urban sprawl, which is characterized by vigorous spatial expansion of urban areas, is a symptom of an economic system gone awry. By transforming pastoral farmland into often-unattractive suburbs, sprawl is thought to disrupt a natural balance between urban and non-urban land uses, leading to a deplorable degradation of the landscape.' This sentiment is often translated into policy through zoning restrictions designed to inhibit the conversion of land from agricultural to urban use (see Bryant and Conklin (1975)). The economist's view of urban expansion stands in stark contrast to this emotionally-charged indictment of sprawl. Economists believe that urban spatial size is determined by an orderly market process which correctly allocates land between urban and agricultural uses. The model underlying this view, which was originally developed by Muth (1969) and Mills (1972) and more completely analyzed by Wheaton (1974), suggests that urban spatial size is determined in a straightforward way by a number of exogenous variables. By showing empirically that urban size is related to the given variables (population, income, agricultural rent, and commuting cost) in the manner predicted by the model, the present paper achieves two goals. First, the empirical results suggest that the economist's view of urban sprawl is justified: rather than being determined by a process which indiscriminately consumes agricultural land, urban sizes are the result of an orderly market equilibrium where competing claims to the land are appropriately balanced.2 Second, by confirming the urban size predictions of the underlying model, the empirical results constitute yet another piece of evidence validating the basic framework of urban economic analysis.3 The plan of the paper is as follows. Section II sketches the structure of the Muth-Mills model and presents the main comparative static results relevant to urban sprawl. With the model's predictions in focus, section III discusses the sample and the data, and section IV presents the empirical results. Section V offers conclusions.

N modern society, where government has assumed a major role in economic affairs and where the electorate has made it increasingly responsible for material well-being, it has become important to analyse the interaction between economic and political systems. Government should no longer be regarded as exogenous to the economic system. This is particularly the case with respect to econometric model building. As some authors have noted, an econometric model may be subject to serious misspecification if an endogenous variable (such as government expenditure) is treated as if it were exogenous.' The study of politico-economic interdependence also has important consequences for forecasting. As the future course of economic events is strongly dependent on government action, existing macroeconometric models that regard government as exogenous are of limited use for prediction. Furthermore, economic policy advice is often unsuccessful because it does not take political repercussions into account. A deflationary policy, for example, will hardly be adopted by a government just before an election because it carries with it a high risk of leading to government's losing the election. Politicoeconometric modelling helps economists concerned with government advising to advance proposals that have a reasonable chance of being put into action. This study puts forward some simple theoretical hypotheses concerning the nature of the interrelationship between the economy and the polity, particularly with respect to (central) government. The basic relationships are reflected in the popularity function, which describes the impact of economic conditions on government popularity; and in the reaction function, which shows how government uses policy instruments to steer the economy in a desired direction. These relationships are econometrically tested with quarterly data for the United States for the period 1953-1975. In the model both voters and government are assumed to be utility maximizers, and government's behavior is restricted by various economic, political and administrative constraints. The analysis shows that the government's (or in the case here dealt with, the president's) popularity is significantly reduced when the rate of unemployment and/or of inflation rises, and that it is significantly increased when the growth rate of private consumption rises. Government reacts to changes in its popularity because this is taken as an indicator of future electoral outcome. When popularity is low, it tries to steer the economy so as to increase its re-election chances; when popularity is high enough, it can afford to pursue ideologicallyoriented policies, which need not always be popular with the electorate. There have been a number of papers that have dealt with the influence of economic variables on election outcomes and on government popularity, most of which are unsatisfactory on theoretical and statistical grounds. There are, on the other hand, only a few that have been concerned with government reaction functions. Moreover, these studies have been either apolitical and interested only in the implied weights of a welfare function (e.g., Friedlaender, 1973); or they have related to only a particular section of the economy (e.g., Received for publication June 14, 1976. Revision accepted for publication November 30, 1976. * University of Zurich. A first version of this paper was written during a stay at the Cowles Foundation, Yale University. It was revised in the light of comments received when it was presented at the Cowles Foundation Seminar and at seminars at Princeton University, the University of North Carolina at Chapel Hill and the Center for Study of Public Choice, Virginia Polytechnic Institute and State University. The authors are especially grateful to A. S. Blinder, J. M. Buchanan, R. C. Fair, G. M. Heal, D. F. Hendry, C. Goodrich, G. H. Kramer, G. Kirchgaessner, D. MacRae, W. D. Nordhaus, W. E. Oates, E. R. Tufte, G. Tullock, R. Wagner, and to the anonymous referees. ' See Crotty (1973), Goldfeld and Blinder (1972), Blinder and Solow (1974, pp. 69-77).

Migration often occurs more than once in an individual's lifetime. Many people may move back to the location where they were born after a stay in another area, or they may move on to yet another new location. In this paper the migrant's location-specific capital and information costs are examined, and empirical findings for the United States are presented and discussed.

A LTHOUGH much research has been done IA-t on the relationships between industrial structure and performance, important puzzles persist. Specifically, it remains unclear whether profits rise with industry concentration when other structural variables, such as market share, are appropriately held constant. Also, what economic phenomena underlie the observed positive profit-market share associations'? This paper seeks to clarify these relationships. Until recently, data limitations have restricted cross-sectional structure-performance analyses to either industry level variables or firm level variables which aggregate quite different activities within a single corporate financial statement.' These limitations are overcome by the Federal Trade Commission's Line of Business survey, which compiles financial statistics disaggregated to the of business (LB) level. A line of refers to a firm's operations in one of 261 manufacturing and 14 nonmanufacturing categories defined by the FTC. The number of LBs per company ranges from I to 47, with an average of 8 lines per company. For each LB, information on pretax profit, advertising, research and development, assets, market share, diversification and vertical integration is reported. When combined with census and input-output data, the FTC line of data allow the estimation of a structure-performance equation of unprecedented richness. A primary emphasis is placed on the theoretical and empirical differences between variables measured at the LB and industry level. To accomplish this task and to relate this paper to the previous literature, regressions are performed at both the LB and industry level.

R ISK transfer and price discovery are two of the major contributions of futures markets to the organization of economic activity (Working (1962), Evans (1978, p. 80), and Silber (1981)). Risk transfer refers to hedgers using futures contracts to shift price risk to others. Price discovery refers to the use of futures prices for pricing cash market transactions (Working (1948), Wiese (1978, p. 87), and Lake (1978, p. 161)). The significance of both contributions depends upon a close relationship between the prices of futures contracts and cash commodities. This paper examines the characteristics of price movements in cash (or spot) markets and futures markets for storable commodities. Section II presents an analytical model of simultaneous price dynamics which suggests that, over short intervals of time, the correlation of price changes is a function of the elasticity of arbitrage between the physical commodity and its counterpart futures contract. Greater elasticity fosters more highly correlated price changes, and thereby facilitates the risk transfer function. The elasticity of supply of arbitrage services is constrained by, among other things, storage and transaction costs. Thus, futures contracts will not, in general, provide perfect risk transfer facilities over short time horizons. The essence of the price discovery function of futures markets hinges on whether new information is reflected first in changed futures prices or in changed cash prices (Hoffman (1932, pp. 258259)). The model in section II provides a framework for analyzing whether one market is dominant in terms of information flows and price discovery. In section III we develop a model based on section II which is appropriate for estimating the lead-lag relationship between cash prices and futures prices. Section IV presents empirical estimates of the parameters of the model for seven different storable commodities: wheat, corn, oats, frozen orange juice concentrates, copper, gold, and silver. The cost of arbitrage between cash and futures differs across these commodities. For this reason we are not surprised to find inter-commodity differences in the correlation of short-run price changes and in the substitutability of futures contracts for cash market positions. With respect to the price discovery function of futures markets, we find that while futures markets dominate cash markets, cash prices do not merely echo futures prices; there are reverse information flows from cash markets to futures markets as well.