II. OVERVIEW OF THE GRADUATE SCHOOLFootnote ⁶

The USDA is a department of the US government that has a long history in the practice of applied statistics and economics.Footnote ⁷ In 1921 Henry Cantwell Wallace was appointed as Secretary of Agriculture, taking over a department already well known for its research in agricultural science, but that had not, under the previous secretary David Houston, encouraged further graduate training for its staff. Wallace, and his new assistant secretary, Elmer D. Ball, were of different minds. Ball, in particular, was concerned about the loss of scientists from the department and wrote to Wallace in April of 1921 suggesting “a system of graduate training within the Department” (Brewster Reference Brewster1985, p. 3). At the same time Wallace himself was concerned to extend the department’s expertise in the area of agricultural economics.

During the summer of 1921, Elmer Ball chaired a committee, including L.C. Gray,Footnote ⁸ then with the Office of Farm Management and Farm Economics, on the formation of a “system of advanced instruction.” The committee did find some precedents, notably a small number of courses offered at the graduate level by the Bureau of Standards. The final memorandum produced by this committee in August of 1921 recommended the provision by the department of opportunities for advanced study in order to improve the efficiency of the department’s work and to aid in the recruitment and retention of the most able people (USDA Graduate School Records, Box 18). It was recommended that instruction be informal and given outside office hours, that the school be self-supporting through the collection of fees, that arrangements be made for the acceptance of credits by other institutions, that cooperative arrangements be made with other institutions in the Washington DC area, and that department regulations be altered to encourage employees to undertake advanced training both at the school and at other universities. It was supposed that students would mostly come from the Department of Agriculture. It was not intended that the school be able to grant degrees, but that a student might fulfill a good part of the requirements for an advanced degree and transfer their credits to a degree-granting university. The committee also recommended a list of possible courses, at both a “general or fundamental” level and an “advanced or graduate” level. The subjects suggested included various areas of agricultural science, statistical methods, and agricultural economics (“History of the Graduate School” 1964, p. 12).

An important related development occurred in 1922 when Wallace appointed Henry C. TaylorFootnote ⁹ as chief of the new Bureau of Agricultural Economics. Taylor also expressed concern about the level of economic and statistical expertise of his staff, and was interested in providing opportunities for further training. Despite the fact that Henry C. Wallace died in 1924, and his successor dismissed Taylor a year later, the BAE continued to grow. By 1929 “the BAE had more social scientists than the rest of the federal government combined” (Banzhaf Reference Banzhaf, Mirowski and Wade Hands2006, p. 14). In terms of economics:

This cluster of statistical and economic expertise is what enabled the Graduate School to provide such high-quality instruction, as many of the instructors held positions with the BAE. Classes were held outside regular office hours, and modest fees were charged to defray costs. The Annual Reports of the Graduate School (USDA Graduate School Records, Box 4) occasionally contain information on the institutions that had given credit for their courses. Initially these tended to be institutions in the Washington DC area; for example, Mordecai Ezekiel transferred his credits to the Robert Brookings Graduate School (headed by Walton H. Hamilton, a leading institutional economist) and completed his PhD there in 1923. Over time the list of universities accepting Graduate School credits expanded to include Columbia, Cornell, Yale, Harvard, Johns Hopkins, Chicago, California, Michigan, New York, Illinois, Iowa, and numerous others. Cooperative arrangements with American University began in 1936 with the AU School of Public Affairs. A short time later American University and the USDA Graduate School were cooperating on a significant number of courses in economics, as well as a few in statistics, with the USDA Graduate School explicitly listing AU courses in its annual announcements of courses and AU giving credit for certain Graduate School courses. This arrangement with American University encountered difficulty in 1940 and ended after the 1944/45 year (Report of the Director, 1943–1944, USDA Graduate School Records, Box 4).Footnote ¹¹

The Graduate School grew in terms of numbers of courses offered and numbers of students, eventually moving well beyond the original vision and becoming by far the most substantial center for training in the federal government outside the armed forces. Training programs were also offered by the Bureau of Standards, the Bureau of the Census, the Department of Commerce, and the Farm Credit Administration, as well as a number of other agencies, but these were small in comparison, and their courses were more focused on specific job skills (Elliott Reference Elliott1940).Footnote ¹² In 1921 the Graduate School offered eleven courses, and the number of courses remained quite small until expanding in the early 1930s and again in the mid-1930s, reaching forty-three in the 1935/36 year, including eight courses in economics, five in mathematics and statistics, and a substantial number of language courses, as well as technical courses in subjects such as photography and glass blowing.Footnote ¹³ The arrival of the New Deal had substantial effects along a number of dimensions. In 1933 Henry A. WallaceFootnote ¹⁴ became Secretary of Agriculture, and Wallace brought with him both a background in agricultural statistics and a fervent belief in the New Deal programs for agriculture. In 1933 the Department of Agriculture had 26,544 employees, but this almost doubled by 1936 and doubled again to 106,217 in 1937.Footnote ¹⁵ The numbers of courses, course enrolments, and the range of instructors available to the school all increased substantially. The extent of the Graduate School’s activity increased dramatically, and this increase was sustained through the 1940s. In 1941/42 year, its twentieth-anniversary year, the Graduate School listed a total of 127 courses offered in the fall and 118 in the spring, as well as 16 courses in the summer. The courses available from American University were in addition to these. In economics, the Graduate School offered an undergraduate course in principles, ten courses described as “Graduate and Advanced Undergraduate” (plus another seven cross-listed with AU), and six “Graduate Courses” (plus three cross-listed with AU). In statistics there were six undergraduate courses listed, four “Graduate and Advanced Undergraduate,” and eleven “Graduate Courses” (plus two cross-listed with AU) (USDA Graduate School Records, Box 1, Catalogs 1921–1956).

The 1940 year saw the establishment of subject-matter departments replacing the original system of subject-matter subcommittees, which had, by then, virtually ceased to exist outside economics. The following eight departments were set up: Biological Sciences (including biology, botany, genetics, bacteriology, plant and animal pathology, and cytology), Physical Sciences (chemistry, physics, meteorology, astronomy, geography, and soil sciences), Mathematics and Statistics, Engineering and Mechanical Arts (architectural and mechanical drawing, engineering, graphic presentation, and glass blowing), Language Aids (foreign languages, English, literature and drama, editing and writing, and speech), Social Sciences (economics, history, sociology, philosophy, psychology, and public law), Public Administration (management and supervision, accounting and auditing, financial administration, administrative law, personnel administration, and public relations), and Office Skills (secretarial training, clerical training, shorthand, office procedures, government correspondence, and business English). This system remained in effect through the period considered here (“History of the Graduate School” 1964, pp. 20–23).

The student body saw related changes. The total number of course registrations began in the 1921/22 year at 319. This number rose to 414 the following year, and then declined to 111 in 1926/27 before starting an upward climb to 505 in 1930/31, then dipping again in the next few years, reaching the low 290s in 1932/33 and 1933/34. From then on, the numbers rose steadily, reaching 7,863 in 1944/45. As these numbers indicate, the student body developed well beyond the Department of Agriculture, and the Graduate School became the provider of further educational opportunities for the federal civil service broadly. This shift began in the mid-1930s as a result of other government agencies asking for permission to let their employees attend classes, and again was accelerated by the demands of the New Deal. The first year in which a breakdown is given is for 1931/32, when 291 course registrations were from members of the USDA and 194 from outside the department. Ten years later there were 1,369 course registrations in the fall and spring terms from within the USDA and 4,858 from other departments and agencies. Employees from the Census, and from the departments of Commerce, Labor, Treasury, Social Security Administration, and many other agencies became well represented.

III. INSTRUCTION IN STATISTICS

For many years the Graduate School offerings in statistics were quite limited (see Table 1). Initially, an introductory statistics course was given and a single advanced statistical methods course. From the fall of 1923 onwards, this advanced course was listed at the graduate level. From 1921 through to the 1925/26 year, Advanced Statistical Methods was taught primarily by Howard Tolley.Footnote ¹⁶ The course covered a full year (two terms) and used Yule’s Introduction to the Theory of Statistics (1911). The course descriptions vary a little, but one term generally dealt with correlation (simple and multiple), curve fitting and applications, and the other term with the theory of sampling and its applications, and/or time series and index numbers. It was from Tolley’s course that Ezekiel first “got started on statistics” (Ezekiel Reference Ezekiel1957, pp. 1–6). In 1926, a few years after completing his PhD at the Brookings Graduate School, Ezekiel returned to the Graduate School as an assistant to Tolley in teaching the advanced course.Footnote ¹⁷ The course was taught jointly by Tolley and Ezekiel for the next three years, and then for one year by C.F. Sarle and E.J. Working.Footnote ¹⁸ At this point the course is described as dealing in the first term with the theory of sampling and its application to the collection and analysis of statistical data (using the text by Yule), and in the second term as covering simple and multiple correlation, both linear and curvilinear, and problems of time series analysis, using Ezekiel’s new book Methods of Correlation Analysis (Ezekiel Reference Ezekiel1930). The course was taught by Ezekiel again in 1931, but only for a term before being taken over by C.M. Purvis. The course content remained much the same, but R.A. Fisher’s Statistical Methods for Research Workers (1928) was used in addition to Yule’s and Ezekiel’s books.

Table 1. Statistics Courses Offered: Selected Years, 1927/28–1936/37 (Graduate Level)

(Source: USDA Graduate School Records, Box 1, Catalogs 1921–1956)

In addition to courses, the Graduate School often played host to visitors who gave lectures or series of lectures. R.A. Fisher gave two lectures at the school on the “Development of the Modern Concepts of Statistical Theory” in the 1930/31 year. Unfortunately, copies of these lectures do not appear to be available in the archives of the school.

This period was also one during which Eziekiel, Tolley, and E.J. Working made important contributions to applied statistics. Working was closely involved with the debates surrounding the statistical estimation of demand relationships (Working Reference Working1927). Tolley and Eziekiel (Reference Tolley and Ezekiel1923) presented the Dolittle method for computing multiple correlations and regressions, and Eziekiel (Reference Ezekiel1924) developed his graphical method for dealing with curvilinear multiple regression problems. Louis BeanFootnote ¹⁹ (of whom more below) developed a shortcut method (Bean Reference Bean1929). Ezekiel’s Reference Ezekiel1930 book Methods of Correlation Analysis was primarily concerned with what would now be called applied regression analysis. According to Fox, “Ezekiel’s book was definitely on the frontier” and “by far the most comprehensive work on applied regression analysis published up to that time” (Fox Reference Fox1989, p. 67). It was clearly Ezekiel’s intent to make the most recent work of “Student” and R.A. Fisher available to his audience, and, to that end, he engaged in correspondence with Fisher to ensure his interpretations were correct (Aldrich Reference Aldrich2000, pp.160–163; 2010). Thus, while the Graduate School was offering only a single graduate-level course in statistics, it was a highly advanced course for its time.

Several things occurred in and around 1933 to change the instructional program in statistics. Edwards DemingFootnote ²⁰ began teaching at the school in 1933, and he offered a new full-year course (in addition to Advanced Statistical Analysis) called Adjustment of Observations. In the first term the course dealt with the theory of errors, Fisher’s fiducial tests of hypotheses respecting parent populations. Bayes’ theorem, Laplace’s generalization, and the posterior method of estimation were applied and compared with fiducial inference. No text was used but papers by Helmert, Gosset (“Student”), R.A. Fisher, Karl Pearson, J. Neyman, Egon Pearson, Shewhart, Keynes, Molina, and Wilkinson were used. The second term treated the subject of least squares and empirical curves from the “general and simplifying view point made possible by recent developments.” Various methods of estimating the parameters of an empirical formula were discussed and some attention given to Karl Pearson’s Chi-square test for goodness of fit (USDA Graduate School Records, Box 1, Announcement, September 1933, p. 6). This course, together with the Advanced Statistical Analysis course, gradually became more elaborate but remained the core of the statistics teaching at the Graduate School through the 1937/38 year. Some other courses were introduced. In 1935/36 Deming offered a course in Advanced Mathematics for Students in Economic Theory and Statistics, covering both linear algebra and calculus, including Taylor series, moments of curves, the Jacobian, and various topics in integration. Later in the same year, an Intermediate Mathematics for Students in Economic Theory and Statistics course was introduced, designed to “move along somewhat slower” than the advanced course. In following years, these courses developed into a substantial series of courses in Mathematical Preparation for Statistics and courses in Differential Equations and Applications. In the 1937/38 year M.A. GirshickFootnote ²¹ began teaching at the Graduate School, offering courses on Multivariate Analysis including “higher algebra” and mathematical statistics.

In September of 1936 R.A. Fisher visited the Graduate School, and gave three lectures on the subject of “Statistical Inference and the Testing of Hypotheses.” In April 1937 J. Neyman visited, and gave a series of lectures and conferences. Neyman’s lectures were on the theory of probability, probability and experimentation, and on the testing of statistical hypotheses. The conferences concerned issues involving field experiments, plant breeding, statistical methods in economic and social research (census by sampling and other problems), time series and related problems in economics, statistical estimation, and an “outline” of the theory of confidence intervals. Both these sets of lectures, and the conferences and discussions, were printed up and published by the Graduate School (Fisher Reference Fisher1936; Neyman Reference Neyman1938a). Neyman’s lectures, in particular, involved issues at the forefront of statistical research at the time, and brought attention to Neyman’s Reference Neyman1934 article on the representative method (Neyman Reference Neyman1934). Issues relating to sampling were just then coming to the forefront of government statistical work due to the 1937 “Check Census of Unemployment” (Duncan and Shelton Reference Duncan and Shelton1978, p. 44). Neyman opened the conference on statistical methods in social and economic research by posing questions about the best method of sampling from a population. Neyman emphatically rejected the method of “purposive selection,” and argued instead for the method of random sampling (Neyman Reference Neyman1938a, pp. 89–90). Those recorded as participating in discussions with Fisher and Neyman included Milton Friedman and Sidney Wilcox, who were then working on a survey of family expenditures.Footnote ²² Friedman’s questions to Neyman involved the issue of ranked data in taking a sample from a full survey. As a direct result, Neyman wrote a paper dealing with this issue, suggesting a solution based on his concept of inductive behavior (Neyman Reference Neyman1938b; see also Friedman Reference Friedman1937).Footnote ²³ The following year two lectures were given by Frank Yates (on the design of factoral experiments and contrasts between the methods of correlation and regression), and a series of four lectures on “The Statistical Method from the Viewpoint of Quality Control” were given by Walter Shewhart.Footnote ²⁴ Later, Deming regularly offered courses on quality control, and this became an area in which he published widely and established an international reputation.

From about 1939–40 the offerings in mathematics and statistics underwent a very considerable expansion. The archives do not provide a specific reason for this, but its timing is coincident with the huge expansion of government statistical work related to the New Deal and, a little later, to the requirements of wartime planning. Large numbers of students came from the Department of Commerce, and the training provided in statistics seems to be designed to accommodate individuals with little previous background. The list of courses for 1940/41 is divided into two categories: mathematical preparation for statistics; and statistics. The former lists a total of fifteen courses, some offered only every two, three, or four years. Of these, six are listed as undergraduate (including basic algebra, calculus, and analytic geometry), and the rest as advanced undergraduate and graduate (including calculus, differential equations, linear algebra, interpolation and finite differences, theory of functions). In statistics seven undergraduate courses (some with multiple sections) and fourteen advanced undergraduate and graduate courses are listed. Again it is clear that there is a rotation of some of these courses over a period of two or three years. Among the advanced courses listed are Graphic Correlation (Bean), Theory of Probability (Girshick), Adjustment of Observations (Deming), Least Squares (Deming), Multivariate Analysis (Girshick), and Analysis of Variance (Girshick). In addition, the Graduate School began offering leading-edge courses on sampling theory and methods (USDA Graduate School Records, Box 1, Educational Courses, 1940–41).

As mentioned above, sampling methods had been experimented with in 1937 with the check census of unemployment, but “the first practical application of probability sampling to a current reporting system in the Federal Government appears to have been the Sample Survey of Unemployment, which was put into operation at the beginning of 1940 by the Works Progress Administration” (Duncan and Shelton Reference Duncan and Shelton1978, p. 47). This survey was conducted by J. Stevens Stock and Lester R. FrankelFootnote ²⁵ (Frankel and Stock Reference Frankel and Stevens Stock1942). Following on from Neyman’s lectures in 1937, the next instruction in sampling methods given at the Graduate School was in the form of two lectures given in October 1938 by Frederick F. Stephan as a part of the course Special Topics in Statistics. This was followed by two lectures from W.G. Cochran in January 1939 and a full-term course, added in the spring of 1940, on Sampling Methods in Social and Economic Data, taught by Stock and Frankel themselves. This was only the second course on sampling to be offered anywhere in the world (Frankel and King Reference Frankel and King1996, p. 70).Footnote ²⁶ The 1940/41 course listings include two courses on sampling, the Stock and Frankel course noted above, and a course on the Theory of Sampling taught by J. Cornfield and W.D. Evans. Sample methods were also introduced into the decennial census in 1940, and Edwards Deming became Mathematical Advisor to the Census Bureau. In the 1941/42 year a course on Theory of Sample Surveys was added, given by William Madow and William Hurwitz,Footnote ²⁷ and from the 1943/44 year Deming organized a regular seminar on sampling. M.H. Hansen also taught along with Madow and Hurwitz, and notes from their courses on sampling became the basis for their 1953 textbook (Hansen, Hurwitz, and Madow Reference Hansen, Hurwitz and Madow1953).

A further formalization of the program in statistics came in the 1942/43 year with the availability of “Certified Statements of Accomplishment in Statistics.” These could be achieved in one of four areas of concentration: processing of data; social sciences; biological and physical sciences; and design and interpretation of sampling surveys. In this particular year, Milton Friedman offered a course on Statistical Analysis of Economic and Social Problems. Also listed is a course on Statistics of Agricultural Economics given by Frederick V. Waugh.Footnote ²⁸ Friedman’s course was described as “Selected topics in the application of statistical methods to demand and cost curves, business cycles, income distribution, measurement of inflationary gaps, etc.” (USDA Graduate School Records, Box 1, Bulletin of Information and Courses 1942–43, p 18). The full program of courses for the Certified Statement of Accomplishment in Statistics in the Social Sciences for the 1942/43 year is reproduced in Appendix 1 (USDA Graduate School Records, Box 1, Bulletin of Information and Courses 1942–43, pp 10–11). Table 2 lists only the graduate-level courses that were a part of the certificate program in the field of social sciences.

Table 2. Statistics Courses Listed for the Certificate Program in Statistics, 1942/43 (Social Science; Graduate Level)

(Source: USDA Graduate School Records, Box 1, Catalogs 1921–1956)

Friedman did not teach at the Graduate School again, but Waugh continued to offer courses in agricultural economics and social welfare issues (see below). Other notables involved in statistics instruction at this time were the biostatistician Jerome Cornfield, later involved in the debate over smoking and cancer, and Solomon Kullback, who became involved in the statistics associated with information theory (the Kullback–Liebler information measure). Overall, the certificate program provided training in mathematics and applied statistics of a type that would have been difficult to obtain at other graduate schools. This program remained largely the same through the rest of the period considered here, although by 1947 Girshick, Stock, and Frankel had all ceased to teach at the Graduate School.

Outside actual course offerings, Harold Hotelling gave a series of lectures in the spring of 1941, but copies of the lectures are not in the archives. The second edition of Ezekiel’s Methods of Correlation Analysis appeared in 1941. Deming and Girshick provided advice on the “more mathematical sections” and especially those dealing with “the sampling significance of results” (Fox Reference Fox1989, p. 67). Between 1943 and 1946, Girshick was working with T. Haavelmo at Cowles on a five equation model of the US demand for food using the limited-information maximum likelihood method (Fox Reference Fox1989, p. 57). Waugh was also on the forefront of statistical research, becoming “the first economist to apply Hotelling’s method of canonical correlation” (Fox Reference Fox1989, p. 69; Waugh Reference Waugh1942).

IV. INSTRUCTION IN ECONOMICS

In terms of economic instruction, the offerings in the early years were both few in number and quite variable. In the first few years, there was a clear focus on agricultural economics. The 1921/22 year saw, in the first term, Henry Taylor teach a course on agricultural economics and, in the second term, a course on the Economics of Commodity Marketing was offered and a course on Agricultural Prices taught by O.C. Stine.Footnote ²⁹ In 1921 and 1922 Mordecai Ezekiel was a student and took agricultural economics from Taylor, and a seminar given by John D. Black on production economics.

The following year (1922/23) saw a division into general and graduate courses with a general-level course, Elementary Agricultural Economics, taught by L.C. Gray, and a graduate course, Advanced Economic Theory, taught by Taylor and C.L. Stewart in the first term, followed by a course on Land Economics in the second term, listed as being taught by Gray and by Richard T. Ely, who was in Washington that term.Footnote ³⁰ The course description provided for Advanced Economic Theory is interesting. It is described as “A review of important economic theories and their application to present day problems” and “A critical study of the principal schools of economic thought, such as the English Classical School, the German Historical School, and the Austrian Psychological School” (USDA Graduate School Records, Box 1, Announcement 1922–23, p. 2). This was clearly a course in general economics rather than one with an agricultural focus. The next year (1923/24) the graduate courses changed completely with a course on Credit, Currency and Prices with Special Reference to the Influence of the Federal Reserve System being offered in the first term by E.A. Goldenweiser, and a course on Economic Cycles offered in the second term by Walter W. Stewart. Stewart had been a colleague and student of Veblen’s at Missouri, and a colleague of Walton Hamilton’s at Amherst before joining the Federal Reserve to develop their Research and Statistics Division in 1922.Footnote ³¹

The next two years show very little economics instruction, with only one half-year course on international trade (with reference to agriculture) actually being offered. A summary view of the graduate-level courses offered in economics in subsequent years is given in Table 3. In the 1926/27 year a course in Price and Price Relationships, taught over a full year by Ezekiel and Louis Bean, was introduced. The first half dealt with Price Movements and Business Cycles and the second with the Economic Basis of Price Analysis, with both parts of the course much concerned with the statistical analysis of price movements. Agricultural prices were a part of this, but the course dealt with a broader range of empirical work on prices and price movements. In 1928/29 Bean’s part of the course was replaced by a term on consumption economics taught by Warren Waite.Footnote ³² This configuration continued into the next year, but with Bean returning with a course on Business Cycles in Relation to Agriculture that he offered in most years until 1933/34 (as a full-year course from 1931/32). The course on Price and Price Relationships was replaced in 1930/31 by a year-long course in Economic Theory taught by E.J. Working and C.F. Sarle. The following year a one-term version of the Price and Price relationships course was taught by E.J. Working, and a one-term seminar in Elementary Mathematical Economics was offered (but withdrawn after a few weeks for lack of students) by Ezekiel. In addition to this, a course in the History of American Agriculture taught by Everett EdwardsFootnote ³³ was introduced in 1927/28. The course focused on social and economic history, and became a staple offering, continuing throughout the period considered in this paper and growing from a one-term to a full-year course.

Table 3. Economics Courses Offered: Selected Years, 1927/28–1936/37 (Graduate Level)

(Source: USDA Graduate School Records, Box 1, Catalogs 1921–1956)

The majority of these courses were taught with a relation to agriculture and agricultural issues, agricultural prices, and agricultural markets, but they were not all focused on agriculture. Many were more empirical than theoretical, and cannot be described as having a neoclassical character. Ezekiel’s course on prices spent time on different methods of arriving at prices, identification issues, issues of time and dynamics, price trends and levels, and seasonal and periodic variations. Waite’s course on consumption started with the social and economic conditions that affect consumption, the effect of changes in income and price on consumption, elasticity, and “changes in consuming habits of the people.” Bean’s course on cycles in relation to agriculture discussed the place of agriculture in various cycle theories, and the relationship between the agricultural and industrial sectors of the economy: the influence of such things as industrial and financial conditions; employment; the purchasing power of “city consumers” on agricultural prices and incomes; and the impact of agricultural conditions on business activity.

The Economic Theory course given by E.J. Working and C.F. Sarle in 1930/31 is particularly interesting as Working was more orientated to neoclassical theory than were Ezekiel or Bean. It is clear the course was intended to be a course in theory, and the course description contains a pitch for the importance of economic theory for the “best work” in applied economics. The description, however, continues as follows:

Ezekiel’s proposed seminar in Mathematical Economics is also of interest, both because he thought of offering such a course at all at that time and because his description indicates both an interest and a concern. The course is described as paying “as much attention to the validity of the assumptions under existing economic conditions as to the theory itself,” while at the same time claiming that mathematical economics is “basic to much of the research work in economic statistics and economic research” (USDA Graduate School Records, Box 1, Announcement, January 1932).

One other Graduate School event should be mentioned. Over the period February and March 1930, the school organized a series called “Special Lectures in Economics.” This included E.B. Wilson talking on “Scientific Method in Economic Research,” John R. Commons on “Evaluating Institutions as a Factor in Economic Change,” Frank Knight on “Fact and Interpretation in Economics,” and John D. Black on “Analytic Methods in Agricultural Economic Research” (USDA Graduate School Records, Box 7, Lectures). Wilson discussed mathematical and statistical methods (sampling, regression, and time series). He clearly felt mathematical methods were necessary for certain complex problems, but also argued that the mathematical method was “likely largely to remain the work of a small fraction of the students in economics” (USDA 1930, p. 4). Commons, of course, stressed the importance of institutions in human behavior, the importance of participation as a method of inquiry into the functioning of institutions, the role of the courts, his concept of reasonable value, and the importance, for the evaluation of agricultural policy, of the distinction between cooperative marketing and collective bargaining. Knight provided his usual critique of scientism and of notions of “social control.” In his lecture, Black distinguished between pure science, specialized pure science, and applied science, and emphasized the significant contributions to be made by agricultural economics in the specialized pure science type (rather than in more applied work). Black argued for a mixture of methods, historical and cross-sectional, quantitative and qualitative, but was somewhat critical of the “institutional approach” for failing to develop and apply more analytical methods to their subject matter:

The instruction offered, then, included consideration of the then-current debates between institutional and more orthodox economics.

Economics instruction over the course of the 1930s expanded significantly, and took on a cast affected both by new developments in economics and the problems and issues of the Depression and the New Deal. The 1933/34 year saw the descriptions of Bean’s course on Business Cycles in Relation to Agriculture make specific reference to the 1929–1933 depression and the efforts made to initiate revival. The course, however, does not appear to have actually been given, and dropped out of the listings in subsequent years. H.J. WadleighFootnote ³⁴ offered a full-year course Theory of Prices and Price Making, the second part of which dealt with theories of imperfect competition, monetary and business cycle theory, and “the institutional school of economics,” as well as “welfare economics and social control.” The following year the course title was changed to Recent Developments in Economic Theory with the focus primarily on new developments in imperfect competition and monetary economics and business cycles, including Keynes’ Treatise on Money. This was clearly a very up-to-date course.

A new course on the History of Economic Thought taught by Max Wasserman was introduced in 1934/35.Footnote ³⁵ The last part of this course covered “recent and modern developments” defined as institutional, quantitative, and welfare economics, and including Veblen, Pigou, J.M. Clark, J.R. Commons, and R.G. Tugwell. Wasserman also taught a course on Inflation in Theory and Practice (USDA Graduate School Records, Box 1, Announcement, August 1934).Footnote ³⁶ The year after (1935/36), Wasserman introduced a Graduate Seminar in Economics and a course in Money and Banking. That year also saw a long series of twenty-two lectures in “Current Economic Problems,” including lectures by Alvin Hansen, Isador Lubin, Calvin Hoover, Jacob Viner, Eveline Burns, Joseph Schumpeter, E.A. Goldenweiser, H. Parker Willis, Edwin Nourse, Morris Cooke, David Saposs, Clarence Ayres,Footnote ³⁷ and Harold Moulton. Outside economics, courses were offered in Administrative Law, and Federal Jurisdiction and Procedure, both containing material on legal questions concerning New Deal legislation. This serves to indicate the impact on the Graduate School of the latest economic thinking and the increasing role of the State in the economy.

The next year saw a vast increase in the offerings in economics, many of which were general in nature. This can be linked to what Andrew Jewett has called “the cultural turn” at the USDA in the 1930s (Jewett Reference Jewett, Allen and Reich2011). The offerings included the Graduate Seminar (Wasserman), History of Economics (Wasserman), Contemporary Economic Problems (Wasserman), History of American Agriculture (Edwards), Economics of Instability (Ezekiel and Bean), Wages and Economic Insecurity (Edward Berman),Footnote ³⁸ The Rise of Modern Industrialism in the US (James Wood), Evolution of Economic Institutions (M.T. Wermel),Footnote ³⁹ Agricultural Prices (F. L. Thomsen),Footnote ⁴⁰ and International Trade (L. Volin). The courses on Trade and the Rise of Modern Industrialism were offered but not given. Courses in Economics in French and Economics in German were offered and given.

Wasserman’s course, Contemporary Economic Problems, dealt with a range of issues, including necessary reforms to the capitalist system, problems of machine industry, agricultural problems, control of industry and trade, unemployment, social security, consumer protection, housing, and wages and hours of work. Wages and Economic Insecurity covered a vast range of issues in labor economics and labor legislation with a distinctly Wisconsin orientation. The course on Economic Institutions is described as examining the fundamental institutions of modern capitalism, “such as Private Property, the Price System, Machine Technique, and Business Enterprise,” a series of topics utilizing a noticeably Veblenian terminology (USDA Graduate School Records, Box 1, Announcement, September 1936). These courses were supplemented by ten lectures on “Frontiers of Law and Economics,” but no details of this series could be found. A lecture series was also given on the USDA and its objectives, including a lecture by R.G. Tugwell on Conservation and Land Use. Courses such as these represent a very explicit consideration of general issues of institutions and institutional reform, which undoubtedly carried with it a positive view of the potential role of State intervention and management of the economy, both in agriculture and beyond.

Many of these courses (or close substitutes for them) were also offered in the following two years, along with Money and Banking (Wasserman), Comparative Current Economic Systems (Berman), and Political Economy and the Law (Joseph Blandi).Footnote ⁴¹ Courses on Trade Unionism and Labor Law (Berman) replaced Wages and Economic Insecurity, with the latter course giving particular attention to the Wagner Act, Social Security legislation, decisions affecting minimum wage laws, and “the controversy over the Supreme Court” (USDA Graduate School Records, Box 1, Announcement, September 1937). In the following year courses on Labor Law and Labor Problems and Policies were offered by David Ziskind,Footnote ⁴² and a course on Current Social Legislation and a course on The Development of English Economic and Governmental Institutions were also offered.

In this period many of the courses had a focus on labor and social legislation, on issues of instability and insecurity, on law and economics, and on institutional development and history. Clearly, a number of these courses were designed with recent New Deal social and labor legislation, and the needs of employees at the Department of Labor, in mind. Instructors such as Ezekiel and Bean were actively involved in the New Deal debates over agricultural policy and the problems of instability more generally. Ezekiel’s work in and around 1938 included work on his Industrial Stabilization Plan, statistical work on the savings behavior of individuals, and his well-known paper on cobweb models and supply adjustment lags as a reason for instability even in agricultural markets characterized by competitive conditions (Ezekiel Reference Ezekiel1937, Reference Ezekiel1938, Reference Ezekiel1939). The significantly institutionalist nature of the economics instruction is quite obvious from the course descriptions. Somewhat in this vein there was also a lecture series, “Democracy,” given in 1938, including lectures by Thurman Arnold, Walton Hamilton, Charles Beard, Ruth Benedict, and Charles Merriam, among many others.

The late 1930s also saw the expansion of cooperation with American University into economics, and this remained a feature until 1944/45. As noted above, the Graduate School organized itself into more formal departments in 1940, and the economics course announcements began listing many courses offered by American University as well as its own. America’s entry into World War II brought about a number of war-training courses in 1942, and the first organized Summer Program that included courses on price regulation and “economic warfare.”

After 1940, however, things began to change in the economics offerings. Questions were asked about the Graduate School, and its legal basis, scope, and philosophy came under attack from Republican Richard N. Elliott, then Acting Controller General (Elliott Reference Elliott1940).Footnote ⁴³ Elliott was concerned with the unclear legal standing of the school, its offering of courses that went “far beyond the strict training of employees for the better performance of their official duties,” and the view of the role of government, consistent with the New Deal, to be found in many of the courses and lectures. One of the outcomes of this seems to have been a refocusing of the Graduate School offerings in economics on agricultural economics. More general courses on economics tended to become those offered by American University (for as long as that relationship lasted). For the 1940/41 year the graduate and advanced undergraduate courses in economics offered by the Graduate School itself were History of Economic Thought, Economics of International Trade, Current Foreign Trade Problems and Policies in Relation to Agriculture, Social and Economic History of American Agriculture, Production Economics, Farm Management, Principles of Agricultural Finance, Economics of Cooperative Organization, Action Programs of the Department, Principles of Land Appraisal, Principles of Insurance, Cotton Marketing, Economies of the Nations of Latin America, and History and Development of Price Policies. David Ziskind did, however, continue to teach a number of courses on labor law, the labor movement, and unemployment.

At the graduate level the school offered Economics of Consumption (Wermel), Economics of Imperfect Competition (Wermel), Forecasting Demand for Agricultural Products (Thomsen), Agricultural Price Analysis (Thomsen), and a seminar on Agriculture under Changing World Conditions (O.C. Stine). Wermel’s course on imperfect competition is interesting as it includes not only Joan Robinson and Edward Chamberlin’s work, but also A.R. Burn’s Decline of Competition (Reference Burns1936) and Walton Hamilton and Associates’ Price and Price Policies (Reference Hamilton1938). In subsequent years this course continued to focus on case studies of pricing in specific industries. His course on consumption is also full of the institutionalist language of planes of consumption, standards of living, consumer organization, and consumer protection. While the courses by Ziskind and Wermel continued to have an institutionalist component, the balance had shifted.

Ezekiel returned to economics instruction in the 1941/42 year with a one-time course on Economic Problems of the Postwar World. At that time Ezekiel was doing a substantial amount of empirical work on saving and consumption (Rutherford and DesRoches Reference Rutherford and Tyler DesRoches2008), and had come to support Keynesian-style expenditure programs. His course outline speaks of giving attention to theories that bear on underutilization of resources and labor, including monopolistic competition and saving-investment equilibrium, and on statistical materials relating to the structural problems of readjustment (USDA Graduate School Records, Box 1, Bulletin of Information and Courses 1941–42, p. 18).

The change in the nature of the Graduate School offerings in economics in the early 1940s can be seen readily in Table 4, which lists the 1942/43 course offerings. As indicated, the economics program developed subsections in economics, agricultural economics, labor economics, and economic theory. In addition, numerous more standard economics courses were cross-listed with AU.

Table 4. Economics Courses Offered 1942/43 (Undergraduate/Graduate and Graduate)

(Source: USDA Graduate School Records, Box 1, Catalogs 1921–1956)

In 1943/44 two new courses began to be offered in agricultural economics: Welfare Aspects of Economic Policy by Waugh and Richard Been; and War and Post-War Food Policy by Waugh and R.H. Allen. These courses contained material relating to Waugh’s own previous work on estimating the marginal utility of money, and his work on food policy, particularly in providing the theoretical rationale for the food stamp plan (Fox Reference Fox1989, p. 69; Waugh Reference Waugh1938).Footnote ⁴⁴ Most of the other courses offered have been mentioned above, but in the economics section, a new course entitled Econometrics: Mathematical and Statistical Analysis of Economic Problems was introduced in the fall of 1944, taught by Been using H.T. Davis’ The Theory of Econometrics as a text. This is the first time the term “econometrics” was used in a course description. Economic theory courses (apart from those offered by AU) were History of Economic Thought (Wasserman), Economics of Imperfect Competition (Wermel), and Post-War Economic Problems (Wermel).

The program offerings in economics continued to grow, but became increasingly specialized. In 1945/46 the economics offerings were divided between a Division of General Economics, including general economic theory, monetary and fiscal, international trade, consumption economics, labor economics, marketing, and research in social sciences; a Division of Economic and Social Policy; a Division of Agricultural Economics; and a Division of International Relations. Ezekiel taught on occasion (courses either on Reconstruction and Full Employment or on Research Methods in Social Science), as did Bean (Measurements of Economic Activity). Gardiner Means taught on Imperfect Competition and Price Regulation starting in 1946, and one can find other institutionalist elements in courses on consumption (Margaret Reid taught at the Graduate School in the late 1940s) and in other areas, but these gradually become less and less characteristic of the offerings as a whole. The courses on business fluctuations moved to become, by 1947, a course on Keynes, Hansen, Robertson, Lange, and the “Stockholm group” (USDA Graduate School Records, Box 1, Bulletin of Information and Courses 1947–48, p. 91).

From the mid-1940s, the focus on agricultural economics increased further. In 1946 the Graduate School introduced a Certified Statement of Accomplishment in Agricultural Economics, and from then on the offerings in economics, other than in agricultural economics, began to decline quite markedly. The requirements for the Certificate in Agricultural Economics are reproduced in Appendix 2. By 1950 only eleven general economics courses (three of them undergraduate) were listed, and not all of these were offered in any given year.Footnote ⁴⁵ This significant narrowing of the offerings of the Graduate School and the increased focus on the core of agricultural economics undoubtedly reflected the political reaction against the New Dealish aspects of the instructional program mentioned above, together with more general attacks on the ideals of planning and the broader social scientific research being carried out by the BAE. The backlash against Carl Taylor’s sociological survey work was particularly strong,Footnote ⁴⁶ but there were many other lines of attack on the policy-making role of the BAE, including its proposed agricultural planning and price policies, and on what was seen as the cadre of urban liberal social scientists in the department (Kirkendall Reference Kirkendall1966). Howard Tolley resigned as chief of the BAE in 1946, and others such as Ezekiel were not far behind. The BAE was eventually abolished in 1953.

V. CONCLUDING REMARKS

The USDA Graduate School clearly was a remarkable organization, and its contributions go far beyond anything discussed here. It began very small with only a few courses, but underwent a massive expansion in the mid- to late 1930s, becoming a unique entity within the federal civil service. The education provided by the school over the period considered here was genuinely at the graduate level. Most studies of the government use of expertise deal with the provision of advice from individuals employed in outside think tanks or research institutions, or who were, at least, trained in the university system. Here we have a case of a government department providing graduate-level training itself, initially for its own employees but ultimately for employees across the civil service. That this occurred in the way it did undoubtedly had something to do with the fact that even prior to the New Deal, the BAE and the Department of Agriculture generally were more aligned to interventionist policies than other departments (for example, Commerce), and were already in the business of training their own people in the statistical and other tools necessary for the implementation of such policies (Hawley Reference Hawley, Furner and Supple1990). The close association of the USDA, the BAE, and the Graduate School with New Deal efforts at planning, however, left them open to attacks from political opponents and business interests, attacks that became increasingly important after 1940.

In terms of statistics, the instruction was often highly advanced, although clearly also applied in orientation. That the school could provide regular instruction by individuals such as Tolley, Ezekiel, Deming, and Frankel, as well as special lectures from R.A. Fisher and Neyman, is historically significant. This level of instruction, available to economists and others in the USDA and elsewhere in the federal government, must have had much to do both with the statistical expertise shown by many of the agricultural economists of the time, and with what Duncan and Shelton (Reference Duncan and Shelton1978) have called the revolution in government statistics that occurred in the 1930s and 1940s. This is perhaps most obvious in the areas of price measures, unemployment measurement, and in the census. It is often forgotten how much was done in statistical research, particularly in applied areas, by statisticians and economists in government employ, working largely outside (although in contact with) the academic world (see also Duncan and Shelton Reference Duncan and Shelton1992). Deming had a particularly important role here due to his previous contacts with both Fisher and Neyman. By the early 1940s the Graduate School was offering an array of statistics courses that many university departments would have been hard pressed to match. The training in sampling, in particular, would have been well in advance of that offered at most universities.

An important observation about the Graduate School is that its statistical sophistication was not wedded to the use of well-specified neoclassical or Keynesian models. Moreover, the important work on sampling, which became such a notable feature of the Graduate School instruction in statistics in 1940, came in good part out of the policy concerns of the New Deal and the related survey work on unemployment and household expenditure conducted by people such as Stock, Frankel, Friedman, Wilcox, and others. This work was underway even before the widespread adoption of Keynesian ideas, and had more to do with the policy concerns and research agenda of the New Deal “planners,” such as Ezekiel and Gardiner Means, than anything else (Stapleford Reference Stapleford2007). The close linking of econometrics to structural economic models that was to become such a feature of the work done at Cowles was not a dominant aspect of the work done at USDA, or of the teaching provided by the Graduate School.

It is difficult to find out who received statistical training from the Graduate School as the archives do not contain lists of students, but Leslie Kish has spoken of the importance of his experiences at the USDA Graduate School in his own development. On Deming he mentions his “unusual ability to get hold of somebody’s important idea and develop it,” particularly in the case of Neyman. Deming invited Neyman to give “his famous lectures at the USDA” and Deming “adopted his philosophy” (Frankel and King 1996, p. 71). It is interesting to note Friedman’s participation in USDA conferences, his involvement with survey work on consumption, and his teaching at the Graduate School. Friedman’s later well-known opposition to the approach of those at Cowles would be in line with the attitudes of others at the school.

In terms of economics, the teaching at the school follows an interesting trajectory. Initially, the economics instruction was limited in quantity and confined very largely to agricultural economics. This period was then followed by one in which the primary courses were the Ezekiel/Bean courses on price and price relationships. These courses went beyond agricultural prices only, were concerned with the general problem of business cycles, and were closely linked up with the instruction in statistics. The courses by Ezekiel and Bean are notable for their very extensive empirical content. In this attribute, they suggest comparisons with Wesley Mitchell’s business cycle course at Columbia or F.C. Mills’ work on both prices and statistics, again at Columbia (Rutherford Reference Rutherford2004). What is especially interesting about Ezekiel and Bean is that they were teaching in both economics and statistics, and that there was very clearly an intimate link between their empirical economic work and their work in statistical methods. In the case of Ezekiel especially we have a case of someone who was in the institutionalist mold, a proponent of economic planning, and who was making contributions both to empirical economics and to applied regression analysis. In this manner, advanced statistical technique coexisted with institutionalist sensibilities. It is true that the work by E.J. Working and, later, Waugh was aligned to neoclassical concepts in demand theory, but Ezekiel and Bean taught regularly in both the statistics and economics programs, something that indicates that the statistical training they offered was felt to be quite consistent with institutionalist economic ideas.

The course offerings in economics then expanded greatly both in number and in breadth in the mid- to late 1930s, clearly in connection with the New Deal. What is especially interesting about this phase of the school’s instruction in economics is how much of it had relatively little to do with agricultural economics as narrowly defined, and the significant part of it that embodied the institutionalist concerns of institutional change and social control. That the USDA Graduate School would offer courses on the history of economic thought, and the evolution of economic institutions, tells one something about the philosophy of those heading up the program (Jewett Reference Jewett, Allen and Reich2011). The scope of the courses offered and their nature reflected more than just the immediate needs of government agencies as created by New Deal legislation. Of course, neoclassical theoretical ideas were by no means absent from the instruction at the school in the 1930s, but there was a clear and substantial institutionalist component, a component that would have been entirely consistent with the favorable attitudes to “social control” then being expressed in government policy.

The economics course offered by the school itself, even at their peak, never amounted to as much of a self-contained graduate program as did the statistics offerings, but together with the courses cross-listed from AU, the Graduate School was certainly offering the equivalent of a good-quality PhD program. To be sure, the Graduate School itself never offered the equivalent of the PhD-level courses in theory as found in schools such as Harvard and Chicago, but it should be remembered that neither Columbia nor Wisconsin offered much in the way of theory instruction until after World War II (Rutherford Reference Rutherford2004, Reference Rutherford2006). The Graduate School’s own courses were certainly up to date, if not as leading edge as some of those in statistics. One very interesting exception to this, however, was Ezekiel’s attempt to offer a mathematical economics course in 1932. At this time such a course would have been very unusual indeed, and to see it being offered by someone of institutionalist background provides a counterexample to the usual view of institutional economists as opposed to more formal mathematical methods.Footnote ⁴⁷

The broader institutionalist component died out of the school’s offerings over the course of the 1940s. Institutionalism was also in retreat in the universities at the same time, but in the case of the main centers of institutionalism, Columbia and Wisconsin, this was largely a matter of the arrival of Keynesian economics and new techniques more aligned to neoclassical approaches. In the case of the USDA Graduate School, what is observed is not a substitution of instruction in these new areas for the more institutionally orientated courses, but a narrowing of the instructional program back towards a focus on agricultural economics, and more in line with the core responsibilities of the USDA. In this case the shift in emphasis seems to have been primarily from the growth of political and business opposition to the previous focus on planning and the relatively statist approach to economic policy matters both in agriculture and beyond. This shift in the economics training provided by the Graduate School undoubtedly played a part in the overall decline of institutional approaches within the federal civil service and the economics profession more generally.