PURPOSE

The COWLES COMMISSION FOR RESEARCH IN ECONOMICS is a not-for-profit corporation, founded in 1932, for the purpose of conducting and encouraging investigations into economic problems. The results of research by members of the Commission's staff are published in two series: Cowles Commission Monographs in book form, and shorter papers, usually reprints of articles from journals, as Cowles Commission Paper, New Series. The Commission is affiliated with the Econometric Society, an international society for the advancement of economic theory in its relation to statistics and mathematics. There is a close relation between the Commission and the University of Chicago.

The main endeavor of the Cowles Commission is to formulate with logical precision economic theories of practical relevance, and to submit them to the test of facts. This task includes the measurement of economic relationships. They can be used to explain business trends and fluctuations, and to estimate in advance the effects of various economic policies of the nation. The required econometric methods (mathematical and, in particular, statistical techniques in the service of economics) are being worked out systematically. At present, they are being tentatively applied to the American economy and its various sectors.

In addition, the Cowles Commission has from time to time studied special subjects of current interest such as the wartime price control (1942–44) and the economics of atomic energy (1946–47), and individual staff members have devoted a part of their time to subjects outside of the general program. One can thus distinguish between the general econometric research program of the Commission and its special studies.

I. ECONOMIC RESEARCH

The statistical testing of economic theories, the measurement of economic relations, and the explanation of business fluctuations observed in the past have an obvious scientific value in themselves. In addition, these studies may prove useful for prediction and policy, especially in that they may provide guidance for steering between inflations and depressions. The meaning of "Economic Structure, Path, Prediction and Policy" will be briefly analyzed in the next pages of this report, where certain other technical terms will also be explained. This logical analysis is followed by the description of the actual econometric work of the Cowles Commission under the following headings: "Adapting Statistical Tools"; "Estimating Structural Relations"; "Revising Economic Fundamentals"; and "Problems for Further Studies."

1. Economic Structure, Path, Policy, and Prediction*

A. Policy and Prediction

1. Knowledge is called useful if it helps to choose the best policy (action).
2. Best policy depends on
   1. the things which one values as goals (e.g., for a firm: profit; for a government: national income, or budget surplus, etc.)
   2. noncontrolled conditions (e.g., for a firm: weather and government policy; for a government: weather)
      The best policy is an action that maximizes (a), given (b).
3. To choose the best policy, it is necessary to predict
   1. the effect of alternative policies under any given noncontrolled conditions;
   2. the future noncontrolled conditions.
4. Thus, all useful knowledge implies prediction; and knowledge useless at one time may become useful later when new goals and conditions present themselves.
5. In human affairs (but also in large parts of technology), to predict is, in general, to estimate, for given conditions and for a given probability level, a probable range of the results of a given policy. This range is wide if the observations are few or subject to large errors, or if structural relations (see B) are subject to large random disturbances.

B. Stochastic Economic Structure and Path

6. The probabilistic character of economic prediction is due to the chance ("stochastic") character of "economic structure." The latter is defined as a set of "structural relations."
7. Each of the structural relations describes either
   1. human behavior (of a specified group of people): e.g., consumers’ demand depends on their current and past income, assets, prices, and on a "random disturbance," the latter being the aggregate effect of numerous, separately insignificant factors; or
   2. technology: e.g., crops depend on acreage, labor, fertilizers, humidity, and a random disturbance; or
   3. legal rules: e.g., price ceilings, tax laws, bank reserve regulations.
      Economic structure is fully described by these relations provided the character of the random disturbances (their variances, covariances) is given.
8. The number of structural relations must equal the number of economic (or nonautonomous) variables. In addition, structural relations contain noneconomic (or autonomous) variables. At any time, the probable range of values that an economic variable can take depends on the following conditions:
   1. the economic structure,
   2. the values of noneconomic variables.
9. Conditions can be noncontrolled (see 2 and 3 above) or controlled. Policies and controlled conditions are identical. Thus a policy fixes either some of the structural relations, or some of the noneconomic variables.
        However, certain government actions may lie outside of government control.
        For example, certain parts of the budget (interest on national debt) are determined by past values of certain variables; and tax rates may reflect political shifts that, in turn, are partly determined by economic conditions at or before election time. Such nonautonomous government actions will not be classified as policies; and relations stating what determines such actions may have to be included among structural relations (possibly using information gathered by political scientists) .On the other hand, government policy may consist in deliberately invalidating such relations of the past and in fixing autonomously variables that were previously governed by structural relations.
        The deliberate introduction of "automatisms" into the economic structure is a particular kind of policy: for example, adopting the legal rule that a change of level of employment and prices by given amounts should be followed by a certain change (stated in advance) in tax rates or public expenditures, with the object of stabilizing employment or prices; or the Bank of England's old rule to raise the discount rate when gold flows out.
10. The path that an economic variable follows through time depends therefore
    1. on the economic structure:
       1. depending on the character of random disturbances, the probable range of deviations from the most probable path will be larger or smaller;
       2. the most probable path may show oscillations if some of the structural relations are dynamic, e.g., if they contain time lags or rates of change or of acceleration;
    2. on noneconomic variables.

C. The Need for Structural Estimation

11. Predictions for a future period, based on observations during a past period, are of different kinds according as whether both or one or none of the types of conditions 8(a), (b) changes within and between the two periods. In particular:
12. If structure is known to remain in the future what it was in the past, and if the noneconomic variables have constant values through both periods, the path of each variable will be predictable from the past, apart from random disturbances. The problem is analogous to that of weather prediction.
13. If structure is known to be retained but the noneconomic variables have assumed and are going to assume changing though known values, it is possible to estimate for each current economic variable its dependence ("regression") on all noneconomic ones and, under certain conditions, on the past ("lagged") economic ones, and to apply this relation to .the future. One can thus estimate the effect of policies that consist in controlling certain variables (tax rates, bank reserve ratios).
14. Finally; if structure is known to change in a given way (example: introduction of price control), the prediction of the effect of this change requires the estimation of the original structure, i.e., of past relations 7(a), (b), (c).
15. Case 14 applies also with regard to the particular policy of introducing "automatisms" (see 9 above) of the most effective kind. For example, to fix in advance the best possible schedule relating tax rates to the unemployment and prices of the previous month, it is necessary to know the lags and elasticities in the consumption equation and in other structural relations at a time when no such legal schedules were in operation. Only when such new device has operated long enough may structural estimation be replaced by the more "mechanical" type of predictions described in 12 and 13.

D. Economic Theory, Statistics, and Mathematics

16. The statistical estimation of the structural relations 7(a), (b), (c) is the "filling of empty boxes of economic theory." The theory is a set of hypotheses. Most of these hypotheses state which variables enter which structural equations, or state certain inequalities (e.g., regarding the signs or relative sizes of certain elasticities) .This is based, essentially, on experience independent of the material that is to be used in estimation. This experience may include statements on rational (i.e., utility-maximizing) behavior and on deviations from it; on a plausible psychology of anticipations; on technological data; etc.
17. Economic theory is useful in the case 14 and useless in the cases 12 and 13. It can be presumed, however, that cases 12 and 13 seldom occur in practice. In particular, any policy that changes one or more of the structural relations of the past gives rise to case 14, and necessitates structural estimation for prediction purposes. Structural estimation may seem useless until a structural change is expected or intended: it comes in very useful then.
         Thus practice requires theory.
18. All the foregoing statements are concerned with the logic of economic knowledge and its uses. This logic is the same whether or not mathematical symbols are used. However, mathematical presentation is of great help in testing the internal consistency of a theory (see 8 on the number of relations and variables); and it is hardly avoidable when the appropriate statistical estimation methods are to be chosen and applied.
19. After stating the hypotheses about the structure, one may find that a certain aggregate of data will permit prediction only in the form of such a wide range of values as to make it useless for policy choice (since widely differing policies will appear to yield equally good results). For this, mathematics cannot be blamed: it will merely reveal what otherwise might remain concealed. Mathematics does not suppress any information available for other methods; and it makes clearer when and how additional information must be used (e.g., extending time series, supplementing them by cross-section data such as attitude surveys, etc.).

2. Adapting Statistical Tools

If economic experiments were possible, estimating the effects of a given policy would not require any new methods of statistical inference. But to use nonexperimental data — such as the time series of economics, or, for that matter, of meteorology — new methods are needed. The need for a revision of statistical methods when applied to economics has been felt, more or less vaguely, for a long time: ever since Marshall pointed out (1890) the "disturbing causes" that make it difficu1t to estimate the elasticity of demand; or at least since Pigou showed (1920) that the positive correlation between the quantity and the price of pig iron calculated by Henry L. Moore could not possibly be interpreted as characterizing a "demand curve," i.e., the behavior of buyers. These "pitfalls in the statistical analysis of demand and supply" (Ragnar Frisch) are due to the nature of economic data that are generated by the interplay of simultaneous structural relations (e.g., demand, or behavior of buyers; supply, or behavior of sellers) and not only by impressed ("autonomous") forces — such as the environment or the experimenter. Traditional statistical methods have to be revised accordingly, and the Cowles Commission has applied itself to the task of revision. The problem of prediction and structural estimation must be studied in the light of the modern theory of statistical inference. This implies formulating economic structural relations in probability terms. The problems involved are indicated by the titles of the following papers, completed during the report period:*

1. "Stochastic Models of Economic Fluctuations" (Hurwicz).
2. "The Probability Approach in Econometrics" (Haavelmo).
3. "Random Simultaneous Equations and the Theory of Production" (Marschak and Andrews).
4. "Statistical Estimation of Simultaneous Economic Relations" (Koopmans).
5. "Estimating Relations from Nonexperimental Observations":
        "The Economist's Problem of Statistical Inference" (Marschak).
        "Sampling Aspects of Structural Estimation and Prediction" (Hurwicz).
        "Iterative Computation Methods in Estimating Simultaneous Relations" (Koopmans and Leipnik).
6. "Methods of Measuring the Marginal Propensity to Consume" (Haavelmo).
7. "Statistical Analysis of Demand for Food: Examples of Simultaneous Estimation of Structural Equations" (Haavelmo and Girshick).
8. "Quantitative Research in Agricultural Economics" (Haavelmo).
9. "Distribution of the Serial Correlation Coefficient in a Circularly Correlated Universe" (Leipnik).
10. "Review of Kendall's book, Contributions to the Study of Oscillatory Time Series" (T.W. Anderson).
11. "Alternative Estimates of Period Length for Oscillatory Time Series" (T.W. Anderson).
12. "Estimation of the Parameters of a Single Stochastic Difference Equation in a Complete System" (T.W. Anderson and Rubin).
13. "Accuracy in the Doolittle Solution" (Leavens).

A collection of articles on "Statistical Inference in Dynamic Economic Models," edited by Koopmans, was completed by the end of 1946 and will be published as Cowles Commission Monograph No. 10. It contains contributions by staff members and by other statisticians who were guests at a Cowles Commission Conference in January, 1945.

A second conference on methods took place in September, 1946, when the following mimeographed papers were discussed:

1. "Multiequational Hypotheses and the Role of Experiment and Decision" (Marschak).
2. "Statistical Models with Disturbances in Equations and/or Disturbances in Variables" (Hurwicz and T.W. Anderson).
3. "Estimation of a Single Equation by Means of the Reduced-Form Method" (T.W. Anderson).
4. "Models Involving a Continuous Time Variable" (Koopmans).
5. "A Sketch of Results on Small-Sample Confidence Regions" (T.W. Anderson, Hurwicz, Rubin).
6. "Remarks on Cowles Commission's Multivariate Problems" (John W. Tukey).
7. "Some Remarks on Multiple Hypothesis" (Rubin).

Although the solution of these methodological problems has advanced far enough to justify experimenting with actual economic hypotheses and data, much remains to be done. We have to learn, in particular, how to deal with more and more general and realistic hypotheses, and how to make a practical compromise between precision on the one hand, and simplicity and flexibility of computational procedures on the other (see below, Section I.5).

3. Constructing Economic Models

For the numerical estimation of relations constituting the economic structure, statistical data are used. Mere contemplation of data is fruitless unless certain hypotheses about the relations are formulated in advance. These hypotheses are suggested by experience other than the particular data to be used in subsequent estimation. Such experience is sometimes called common sense; it is also contained in technological and legal data, or provided by earlier and independent historical or even statistical studies. A part of this experience is expressed in economic theories that are deduced from the fact that certain types of human behavior are rather plausible. However, of the existing writings on modern economic structure — including the literature on the production and distribution of national output, on business cycles, unemployment, or international trade — very few contain clear, logically consistent, and complete hypotheses on structural relations. Yet such hypotheses are necessary before a meaningful analysis can be attempted. We have tried to scrutinize much of the useful existing literature: the major propositions of the modern theory of the firm, the consumer's behavior, the behavior in the money market, etc. The macroeconomic models constructed in the 1930s by Frisch, Kalecki, Tinbergen, and others, the mathematical restatement of certain aspects of Keynes' theories by Hicks and Lange have been helpful in starting our work. Possibly the simplest model is the one discussed, as an illustration of the method, in Haavelmo's paper on "Methods of Measuring the Marginal Propensity to Consume" (quoted above, item 6). This model involves only one structural relation, viz., the relation describing consumers' saving as a function of their current and past income (after taxes) , while the entrepreneur's investment and the government receipts and expenditures are considered as autonomous. Owing to the sharper formulation of the hypothesis and to the use of appropriate estimation methods, Haavelmo's results may signify a progress compared with discussions based on similar models in recent years. Yet, realism requires more complicated models. For example, a distinction must be made between houses and other consumers' goods and, possibly, between owner-occupied and rented dwellings, because of different motivations involved in each case. Private investment, instead of being autonomous, may prove to depend on current or past prices, wages, or interest rates. Separate motivations may rule the expansion of plant and the changes of inventories. The behavior of borrowers and lenders must be inserted into the system. Government revenue must be broken down into its factors, e.g., tax rates and the taxable income. And so on. Accordingly" a series of more detailed alternative tentative models (each involving a larger number of relations) has been constructed by Klein. To estimate these relations in a provisional way, he first used the traditional method of regression (in item 16 below) .In the meantime, other staff members achieved progress with the revision of statistical methods discussed in Section I.2 above. New computational techniques were then tried out, and revised estimates for economic structural relations were prepared. The economic hypotheses themselves are being constantly revised; to use them for forecasting the near future is a way to test them. Much more work will thus be required to get firm ground.

Besides the analysis of the economy as a whole in terms of major aggregates — national income, investment, price level, etc. — there is a need, both scientific and practical, for the analysis of special sectors of the economy. Haavelmo's models for agriculture have been listed above (items 7 and 8, in Section I.2) .In these models the behavior of sellers and of buyers in the wholesale and retail food markets are stated and estimated separately, while all other structural relations of the economy are represented by a single relation (which they imply) between current and past national income and certain autonomous variables. This work of Haavelmo was a joint project of the Cowles Commission and the Agricultural Economics Research group of the University of Chicago (D. Gale Johnson, William H. Nicholls, and Theodore W. Schultz); it owes much to the cooperation with Girshick and to the general support of James Cavin of the U.S. Bureau of Agricultural Economics.

On similar lines, Patinkin attempts a sectional analysis of manufacturing industries.* This implies in particular a critique of theories of the behavior of entrepreneurs (profit-maximizing or not, competitive or not) that determines output, equipment, inventories, etc. Very likely, time series of national aggregates will have to be supplemented by business records of single firms, and possibly by more intensive sampling of businessmen's attitudes. The practical difficulties that these methods involve may, however, be very great.

Another sectional study, by Sonia Adelson, is devoted to international trade. The foreign trade balance and the international terms of trade, treated as autonomous for purposes of other models, have themselves to be explained. One needs a simplified model of structural relations, which would involve the economies of more than one country but would not treat them in much detail. Jacob Hartog, a Rockefeller Fellow and guest of the Cowles Commission, has also been associated with this subject.

The titles of the following papers, completed during the period covered by the report, indicate the content of our work devoted to the criticism, construction and use of economic models:*

14. "Cross Section of Business Cycle Discussion" (Marschak).
15. "Multiplier Effects of a Balanced Budget" (Haavelmo).
16. "Economic Fluctuations in the U.S. 1920–1941" (Klein), mimeographed. 67 pages. Abstract of this paper was presented at the meeting of Econometric Society, on January 24, 1946, in Cleveland (see Econometrica, Vol. 14, April 1946, pp. 159–162).
17. "A Post-Mortem on Transition Predictions of National Income" (Klein).
18. "Forecasting National Product and Employment" (Klein).
19. "The Use of Econometric Models as a Guide to Economic Policy" (Klein).
20. "Market Adjustment and Inventory Equations" (Patinkin), mimeographed.
21. "Preliminary Model for Manufacturing in the U.S." (Patinkin), mimeographed.

4. Revising Economic Fundamentals

In postulating structural economic relations that have to be submitted to the statistical procedures of testing or estimation, one possibility is to accept, as a first approximation, the assumption of rational economic theory: that people behave so as to maximize their profits (or, more generally, their satisfaction) according to the best knowledge they have. When trying to make use of theory for econometric purposes, one soon discovers, however, that much clarification is required to give the hypotheses a form amenable to statistical tests. In particular, behavior in conditions of uncertainty has only just begun to be analyzed by students of investment. What is the rational behavior for a firm that has only imperfect information, e.g., only a finite sample, from which to estimate the properties of its future product or of its future market? Or, if the firm does not use its imperfect information in the way that the rational statistician would advise it to follow, what are the most plausible psychological patterns according to which modern man's expectations of the future are based on current and past information? What behavior can be predicted when a seller faces only a small number of competitors and buyers? Finally: given a theory of behavior (rational or not) of single individuals in the markets for single commodities, what is the best method of combining these innumerable atomistic structural relations into a useful system of relations between national aggregates — total income, savings, profits, employment, price and wage level, interest rate, etc. Such a system must be both manageable and accurate enough to be applied in prediction and policy decisions. Inasmuch as construction of economic models and estimation of structural relations (Section I.3) has so far proceeded without waiting for complete answers to fundamental problems of economic behavior, the work has been on a somewhat intuitive basis. To improve our theoretical postulates means revising basic economics. The type of problems in question is indicated by the titles of the following papers:*

22. "The Theory of Economic Behavior" (Hurwicz).
23. "Neumann's and Morgenstern's New Approach to Static Economics" (Marschak).
24. "Macroeconomics and the Theory of Rational Behavior" (Klein).
25. "Theory of the Firm and of Investment" (Hurwicz).
26. "Remarks on the Theory of Aggregation" (Klein).

5. Problems for Further Study

The adaptation of statistical tools to econometric analysis will require further work. More data will be needed to reduce the range of error in estimates. In particular, in order to add information contained in quarterly (and perhaps monthly) instead of annual data, we have to find suitable methods of treating seasonal effects and of allowing for correlation between the random disturbances occurring in successive months or quarters. Furthermore, in many cases it is possible, or even necessary, to combine time series with "cross-section" data (e.g., surveys of households or firms). Finally, constant criticism of past estimates in the light of recent data will be extremely useful.

Optimal methods should extract from the data the maximum amount of relevant information that can be obtained through computationally feasible procedures. Computational labor can be reduced if only those structural relations are estimated that are deemed relevant to specific policies under discussion; yet the loss of precision in so doing is to be investigated. To increase the amount of information gained by a given amount of effort we shall also have to devise flexible methods of computation, so that several alternative hypotheses can be used simultaneously. It has been stated above (Section I.3) that estimation of structural relations requires a certain minimum of hypotheses that are not verifiable from the particular data used for such estimation. However, hypotheses additional to this minimum are amenable to statistical test; at the same time they contribute to the precision of estimates. A statistical theory should be developed that would guide the selection of such additional hypotheses.

The type of hypotheses used has been limited so far to simple cases. The structural relations were assumed linear. The probabilistic character of the estimates was assumed to be due to (additive) random disturbances in relations, while observation errors in variables have been neglected (other investigators have often made the opposite assumption). These assumptions will have to be relaxed or modified in due course. Finally, if known methods are applied to structural relations that involve time lags, estimation biases arise that are negligible only in larger samples than those available at present in the form of time series of annual economic observations (20 years or so). The sampling theory related to "medium-size" time series is a matter for further study.

The revision of economic theories and construction of economic models also requires further study of a number of fundamental problems. What are the optimal methods of aggregation — i.e., the most suitable techniques of forming averages ("index numbers") which would bring the theory of single firms and households to bear upon the political economy of a nation? How should rational economics of the behavior of single firms and households be developed or corrected to become realistic in conditions of uncertainty? What is the size of error committed in treating all governmental actions (e.g., the setting of tax rates) as independent of economic conditions; or, alternatively, how can the dependence of such actions on economic conditions be described in quantitative terms?

As indicated, it will, in general, not be necessary to estimate the whole system of structural relations in order to estimate the results of policies under discussion. To arrange the economic and statistical studies in the order of practical priorities, a systematic classification of policies will be necessary. Policies can be classified according to goals (e.g., high income, stable income, income equalization) or according to formal criteria: some policies consist in discarding certain structural relations and substituting others; some policies merely change the numerical value of certain (nonobserved) structural parameters such as the elasticities of demand or of production; still others change the (observed) values of autonomous variables.

Still further problems loom on the horizon. While trying to explain, or to devise possible controls of, business fluctuations one should not forget the secular trends and the long-run effects of technological, demographic, or institutional changes. In fact, business-cycle policies may be regarded as a special case of general welfare economics which aims at the goal of high employment together with the goal of the best possible utilization of resources. While some thinking on these lines is being done within the Cowles Commission at present, especially in connection with problems of inventions and of optimal location, it is too early to give a more detailed statement on progress in this field of work.

II. SPECIAL STUDIES*

Stock Market Forecasting. In a paper of this title, Cowles continued the investigations carried out in the past,* analyzing the degree of success of stock forecasting agencies and testing the existence of regularities in the market.

Applications of General-Equilibrium Theory. In two monographs (No. 7 and No. 8) by Mosak and by Lange, the general-equilibrium theory, and especially the modern analysis of stability and of rival and complementary goods, is made fruitful in two fields: the theory of international trade and the theory of money (and other assets) and employment. Lange also dealt with related subjects in several articles.

Price Control and Business. A monograph (No. 9) by Katona summarizes the results of extensive field studies conducted in 1942–44 by a staff of assistants under his direction and with the sponsorship of the Rockefeller Foundation and of the Price Conference of the National Bureau of Economic Research. Producers and distributors of consumers' goods in the Chicago area were interviewed by the "open-end question" method, and their attitudes toward price control recorded and analyzed. The study has probably contributed to the fund of experience relating to the use of interview methods for economic research (after leaving the Cowles Commission the author has continued to develop and apply the "open-end question" method for economic surveys on behalf of other institutions).

Transportation and Location. Koopmans is engaged in a study of the economics of transportation, especially shipping and railroads. The marginal cost of transportation on a given route is derived from a study of the most economic routing of empty equipment from points of discharge to points of loading. Principles are formulated for a system of freight rates expressing marginal cost. The optimal locational distribution of industry in a country requires a rate system of this kind, so as to induce a more balanced use of the transportation equipment than can be obtained through rates depending on distance only.

Silver. Leavens continued to collect material on the monetary use and general economics of silver and published articles on the subject.

Econometrics and Ancient History. Davis continued his work on construction and interpretation of price series for Alexandria and for the late Roman period.

Mathematical Tables. Davis continued his work in the preparation of tables of mathematical functions; he also studied the history of these functions and reviewed a number of existing tables.

Economic Effects of Atomic Energy as a Source of Power. This exploratory study was begun in October 1946, and is sponsored by the Social Science Research Council (Committee on Social Aspects of Atomic Energy, under the Chairmanship of Winfield W. Riefler) and the Rockefeller Foundation. It is intended to list the possible effects that moderately priced electricity produced from a fuel with negligible transportation costs may have on world economy. Codirectors of the study are Marschak and Schurr, assisted by Boorstein and Perazich.

III. RESEARCH PERSONNEL

Since 1945 the duties of research personnel have been redefined as follows: research associates devote most of their time to the work with the Cowles Commission, aided by full-time or half-time research assistants. Research consultants cooperate in the work of the Commission by participating in staff meetings, by correspondence, or by other occasional contributions. Only active members will be kept at any time on the personnel list such as the list on the cover of this report.

Some of the staff members are members of the Department of Economics. Certain other staff members have been granted the rank of assistant professor or instructor at the University of Chicago by the University's Board of Trustees.

The biographies and brief surveys of activities of research staff are given in Section XII.

IV. SARAH FRANCES HUTCHINSON COWLES FELLOWSHIPS

Two graduate Fellowships for women were awarded by the University of Chicago upon nomination by the Cowles Commission. Applicants were required to be preparing for the degree of Master of Arts or Doctor in the fields of Social Sciences and Statistics, with preference to be given to students who would be working on quantitative economics or mathematical statistics.

The Fellowships were awarded for the first time in 1946–47 to Sonia Adelson and Selma Schweitzer.

V. COWLES COMMISSION SEMINARS

Seminars for faculty members, graduate students, and others interested in the studies of the Cowles Commission were started in July, 1943. The following papers were presented and discussed:

VI. STATISTICAL AND ECONOMETRIC TEACHING
IN THE DEPARTMENT OF ECONOMICS OF THE UNIVERSITY OF CHICAGO

Staff members of the Cowles Commission participate in the teaching activities of the University of Chicago, especially in the field of statistics and economic theory .Stress is laid upon the connection among mathematics, economics, and statistics. The following courses are listed (but not all are given every year):

Introduction to Statistics. Elementary principles of statistics. Main topics: frequency distributions, averages, dispersion and skewness, time series, index numbers, simple correlation, elements of sampling, and statistical inference.

Statistical Inference (two two-course sequences). A survey of the principles of statistical inference, with emphasis on the techniques useful in applying these principles to the analysis of social, economic, and business data. Among the subjects treated will be: elements of probability; concepts of population, sample, and sampling distribution; choice of estimates in the light of their sampling properties; testing hypotheses with reference to specified alternatives; principles of sampling and sample design; analysis of proportions, means and standard deviations; simple, partial, and multiple regression and correlation. These courses are given in two sequences covering substantially the same statistical principles and procedures, one section making less use of explicit mathematical formulations than the other.

Introduction to Econometrics. Statistical testing of economic theories. Numerical estimation of demand and cost functions and other functions occurring in the theory of the firm and household, the theory of markets, and the theory of national income. Estimation of economic models. Statistical prediction under conditions of changing economic structure and policy.

Statistical Methods of Measuring Economic Relations. Systems of economic relationships. Prediction and structural estimation. Criteria for identifiability and methods of estimation, of economic relations. Problems connected with time lags.

Applications of Statistical Analysis. The content of this course will vary from one year to another, but will concern applications of statistics to business, economic, and social data. For example, members of the class individually or jointly may carry tl1rough a statistical investigation; or a series of statistical studies in economics or business may be analyzed in detail; or some special field of application may be studied.

The Main Economic Magnitudes. Survey of sources and methods for estimating national income, capital formation, consumption, balance of payments, monetary circulation, and prices. Attention is given to practical work. Students have opportunity to familiarize themselves with the sources and techniques relevant to the statistical study of the American economy.

Introduction to Mathematics for Economists. A survey of those parts of mathematical analysis which are used in economics. Fundamental mathematical concepts: function, limits. Dimensions. Elementary Calculus. The application to economics is stressed throughout.

Problems in Mathematical Economics. Elements of advanced calculus and of ordinary and differential equations applied to fundamental economic problems. The material is arranged in the order of increasing mathematical difficulty.

Econometrics of Business Fluctuations. Mathematical formulation and statistical testing of theories of economic change. Growth and fluctuations. Prediction and Policy.

These courses are additional to the relevant courses given in the School of Business and the Departments of Education, Mathematics, Psychology, Sociology, and Zoology.

Staff members of the Cowles Commission are scheduled to give a large part of the courses listed above and of the following courses: The Divisional Course in Economics; and The Theory of Income and Employment.

VII. ADVISORY COMMITTEE

The University of Chicago Advisory Committee of the Cowles Commission for Research in Economics meets from time to time to coordinate the work of the Cowles Commission with other research and teaching work of the University. The chairman is Theodore W. Schu1tz, chairman of the Department of Economics. The vice-chairman is Louis Wirth, Secretary of the Social Science Research Committee of the University.

VIII. GRANTS

Acknowledgment is made to the Rockefeller Foundation and the Social Science Research Committee of the University of Chicago for financial assistance in the research work of the Cowles Commission. Several of the research staff have held Fellowships from the Guggenheim Foundation, the Social Science Research Council, or the University of Chicago.

IX. OFFICES, EQUIPMENT, AND LIBRARY

The Cowles Commission occupies rooms provided by the University of Chicago in its Social Science Building, 1126 East 59th Street. There are 6 individual offices, a large room with desks for staff members, also used for staff conferences; a room for secretarial and computing staff, provided with 7 adding and calculating machines, 4 typewriters, and other equipment; and two rooms for library.

The Commission has been building up a library containing most of the material pertinent to its investigations and now has some 1800 books, 3100 pamphlets, and 350 bound volumes of periodicals. In addition, the library of the late Professor Henry Schultz, belonging to the University, is kept in the Commission's offices; it contains 950 books and 1750 pamphlets. About 115 periodicals are received regularly. The library is open to advanced students by arrangement.

X. THE ECONOMETRIC SOCIETY

The Cowles Commission since its organization in 1932 has been affiliated with the Econometric Society, an international society for the advancement of economic theory in its relation to statistics and mathematics. The Society's offices have been located at the headquarters of the Cowles Commission in Colorado Springs and later in Chicago, and its quarterly journal, Econometrica, has been published there. Membership in the Society has grown from 163 in 1931 to 721 in 1941 and 774 in 1946. Nonmember subscribers to Econometrica, chiefly libraries, have increased from 77 in 1933, its first year of publication, to 295 in 1941 and 445 in 1946. Members and subscribers are located in 51 different countries. During the war the office could not communicate with those in many of the belligerent countries, but contact has now been renewed with a good proportion of them. Through 1946, 14 volumes of Econometrica, totalling about 5500 pages, have been published.

During the last five years members of the Cowles Commission staff have continued to take an active part in the work of the Society. Marschak served as vice-president in 1944 and 1945 and as president in 1946. Cowles continued as secretary and treasurer throughout the period. Lange was acting editor of Econometrica from 1943 to 1945, when it was impossible to communicate with the editor, Ragnar Frisch, of the University of Norway. The following continued to serve during the period covered by this report: Davis, associate editor; Marschak, member of the advisory editorial board; Leavens, managing editor; and Cowles, business manager. Koopmans, Lange, Leavens, and Marschak have served on the program committee for meetings of the Society.

During the war meetings had to be suspended for a time, but one was held at Cleveland, Ohio, September 13–15, 1944, in conjunction with the meeting of the American Association for the Advancement of Science; and another in the same city, January 24–27, 1946, in conjunction with meetings of the American Economic Association, the American Statistical Association, and the Institute of Mathematical Statistics.

XI. COWLES COMMISSION PUBLICATIONS

1. Cowles Commission Monographs

See Complete LIST OF MONOGRAPHS

2. Cowles Commission Papers, New Series

At the end of 1943 the policy was adopted of having reprints of papers by members of the Commission's research staff bound in special covers as Cowles Commission Papers, New Series (available on line).

Next >>